Você está na página 1de 403

Con tents

Pr eface


1-1 Bas ic Defin itions 13
1-2 Con version of El ectric Energy by the Tr an sforme r 15
El ectromechanical Energy Con versio n by an
Electrical Machin e 18
1-4 Functiona l Classifi cation of Elect romagnetic
En er gy Conve r t ing Devices 24










An Outline of 'I'ransiormers
P urpose, Applications, Ratings 27
Const ru ction of a Transformer 31
Elec tromagne tic Processes in the Transformer
a t No-Load
Th e No-Load Condition 43
Voltage Equations 45
Vari ations in EMF with Time. An EMF Eq uation 46
Th e Magnetization Curve of the Transforme r
Th e No-Load Cur ren t W aveform 49
T ra nsformer Eq ua tio ns a t No-Lo ad in Compl ex
Form .50
No-Loa d Losses 52
The Effect of the Core Loss on t he T ransformer 's
Pe rform ance at No-Loa d 53




El ectromagnetic Pro cesses in the Transformer on

The Magnetic Field in a T ran sform er on Load.
Th e MMF Equati on. Th e Leakage Inductance
of th e Windings 56
Voltage E qua ti ons of the Tr an sform er Windings 60
Transferring t he Secondar y Quan tities to t he
Primary Sid e 6?
. .




The Phasor Diagram of a Transformer 65

The Equivalent Circuit of the Transformer 68
The Per-Unit Notation 69
The Effect of Load Variations on the Transformer 72
Energy Conversion in a Loaded Transformer 75
Transformation of Three-Phase Currents and
Methods of Three-Phase Transformation. Winding
Connections 79
A Three-Phase Transformer on a Balanced Load 83
Phase Displacement Reference Numbers 84
The Behaviour of a Three-Phase Transformer
During Magnetic Field Formation 89



Measurement of Transformer Quantities

The Open-Circuit (No-Load) Test 99
The Short-Circuit Test 102



Transformer Performance on Load

Simplified Transformer Equations and Equivalent Circuit for 11 1 0 106
Transformer Voltage Regulation 107
Variations in Transformer Efficiency on Load 111









Tap Changing
Off-Load Tap Changing 113
On-Load Tap Changing 114


Calculation of Transformer Parameters

No-Load (Open-Circuit) Current and Mutual Impedance 117
Short-Circuit Impedance 119


Relationship Between Transformer Quantities and

Variations in the Voltage, Current, Power and
Mass of a Transformer with Size 121
Transformer Losses and Parameters as Functions
of Size 123



Chapter 10

Multiwinding Transformers. Autotransformers

Multiwinding Transformers 125
Autotransformers 133

Chapter 11

Transformers in Parallel
Use of Transformers in Parallel 138
Procedure for Bringing Transformers in for Parallel Operation 139
Circulating Currents due to a Difference in
Transformation Ratio 141
Load Sharing Between Transformers in Parallel 14:/



Chapter 12

Three-Phase Transformers Under Unbalanced Load 145

Causes of Load Unbalance 145
Transformation of Unbalanced Currents 146
Magnetic Fluxes and EMFs under Unbalanced
Load Conditions 151
Dissymmetry of the Primary Phase Voltages under Unbalanced Load 154
Dissymmetry of the Secondary Voltages under
Unbalanced Load 156
Measurement of the ZPS Secondary Impedance 160
Single- and Two-Phase Unbalanced Loads 161

Chapter 13

Transients in Transformers
Transients at Switch-On 164
Transients on a Short-Circuit Across the Secondary Terminals 167


Chapter 14

Overvoltage Transients in Transformers

Causes of Overvoltages 171
The Differential Equation for the Initial Voltage Distribution in the Transformer Winding 172
Voltage Distribution over the Winding and Its
Equalization 175


Special-Purpose Transformers
General 177
Three-Phase Transformation with Two Transformers 177
Frequency-Conversion Transformers 178
Variable-Voltage Transformers 179
Arc Welding Transformers 180
Insulation Testing Transformers 181
Peaking Transformers 182
Instrument Transformers 182


Heating and Cooling of Transformers

Temperature Limits for Transformer Parts under
Steady-State and Transient Conditions 184
Transformer Cooling Systems 186


Chapter 15
Chapter 16
Chapter 17



Transformers of Soviet Manufacture
USSR State Standards Covering Transformers 189
Type Designations of Soviet-made Transformers 190
Some of Transformer Applications 191
A general theory of electromechanical
energy conversion by electrical






Chapter 19
Chapter 20



Chapter 22
Chapter 23

Classification of Electrical Machines 192

Mathematical Description of Electromechanical
Energy Conversion by Electrical Machines 195
Production of a Periodically Varying Magnetic
Field in Electrical Machines
A Necessary Condition for Electromechanical
Energy Conversion 201
The Cylindrical (Drum) Heteropolar Winding 202
The Toroidal Heteropolar Winding 206
The Ring Winding and a Claw-Shaped Core 206
The Homopolar Ring Winding and a Toothed
Core 206
Basic Machine Designs
Modifications in Design 207
Machines with One Winding on the Stator and
One Winding on the Rotor 211
Machines with One Winding on the Stator and
Toothed Rotor and Stator Cores (Reluctance
Machines) 214
Machines with Two Windings on the Stator and
Toothed Cores for the Stator and Rotor (Inductor
Machines) 218
Conditions for Unidirectional Energy Conversion
by Electrical Machines
The Single-Winding Machine 227
Two-Winding Machines 230



Windings for A. C. Machines

Introductory Notes 235
The Structure of a Polyphase Two-Layer Winding 235
Connection of Coils in a Lap Winding. The Number of Paths and Turns per Phase 240
Coil Connection in the Wave Winding 244
The Selection of a Winding Type and Winding
Characteristics 246
A Two-Pole Model of a Winding. Electrical
Angles between Winding Elements 247
Two-Layer, Fractional-Slot Windings 250
Field Windings 255
Calculation <if the Magnetic Field in an Electrical Machine
The Statement of the Problem 257
Assumptions Made in Calculating the Magnetic
Field 260
The Spatial Pattern of the Magnetic Field Set Up
by a Polyphase Winding 262
Calculation of the Mutual Magnetic Field Ior [J
Polyphase Winding 264
Effective Length of the Core $6ff


Chapter 24
Chapter 25
Chapter 26
Chapter 27

Cha pter 28

The Mutual Magnetic Field of a Phase W inding

and Its Elements
The Magneti c Field and MMF due to a Basic Set of
Currents 267
The Effe ct of Core Sali ency. The Carter Coefficien t 270
The MMF due to a Basic Coil Set 272
Expansion of th e Periodic MMF du e to a Basic
Coil Set into a Fourier Series. Th e Pitch Factor 276
The Phase MMF. The Distribution Fa ctor 280
Pulsating Harmonics of the Phase MMF 287
The Mutual

Magnetic Field of a Polyphase


Presentation of th e Pu lsating Harmonics of the

Phase MMF as the Sum of Rotating MMFs 288
Presentation of Pha se MMF H armonics as Complex Time-Space Functions 291
Time and Spa ce-Time' Complex Quantities and
Functions of the Quantities Involv ed in Opera tion of a Polyphase Machine 294
Th e MMF of a Pol yph ase W inding , Its R otating
Harmonics 297
The Fundamental Component of the Magn etic
Flux Density in a Polyphase Winding (the Rotating Field) 303
Magn etic Flux Density Harmonics in th e Rotat ing Magneti c Fi eld of a Pol yphase Winding 306
The Magnetic Field of a Ro lating Field Winding
The Magneti c Field of a Concentrated Fie ld
Winding 316
Th e Magnetic Field of a Distributed Fie ld
Winding 319
The Rotatin g Harmonics of th e Excitation Fie ld 321

Flux Linkages of an d EMFs Induced by Rotati ng

F ields
Introductory Not es 323
Th e Flux Linkage and EMF of a Coil 323
Th e Flux Linkage and EMF of a Coil Group 328
The Flux Linkage and EMF of a Phase 330
Th e Flux Linkages and EMFs of a Pol yphase
Winding . A Space-Time Diagram of Flux Lin ka ges an d EMFs 333
Th e Flux Linkages and ' EMFs du e to th e
Harmonics of a Nonsinusoidal Rotating Magnetic Fielrj 3/15
The In du ctances of Polyphase Win dings
Th e Useful Fi eld and the Leaka ge Field 341
The Main Self-Induc tan ce of a Phase 342
The Main Mutual Inductance Between the Phases 343
The Main Mutual Inductance Between a Stator
Ph ase and a Ro tor Phase 341



Chapter 29

Th e Main Self-Inductance of the Complet e

Winding 345
Th e Main Mutual Induct ance betwe en a Primar y
Ph ase and th e Second ar y Wi n ding 347
Th e Leakage Inductance of th e Complete W inding 348
Thc Elcctromagnctic Torque
The Torque E xpressed in Ter ms of Vari ation s in
th e Energy of th e Magnetic Field 351
Th e El ectromagneti c Torque Expressed in Terms
of El ectroma gnetic Forces 358
Electromagnetic Force Distribution in a Wound
Slot 367


Chapter 30

Energy Conversion by a Hota ti ng Magnetic Field

El ectromagnetic, El ect ri c and Magnetic Power 372
Energy Conversi on in an El ectrical Machine and
Its Model 376

Chap ter 31

Energy Conversion Losses and Efficiency

Introductory Notes 379
Electrica l Losses 380
Magnetic Losses 387
Mechanical Los ses 396


Bi bliog raphy





The subject matter in the text is presented in the sequence

traditionally followed in the Soviet Union. It starts with
transformers, passes on to induction and synchronous
machines, d.c. machines, and concludes with a.c, commutator machines. Separate chapters are devoted to a general
theory of electrical machines, machine design and engineering, and transients in electrical machinery.
The electromagnetic processes that take place in electrical
machines are examined from the view-point of electromechanical and mechanoelectrical energy conversion . With
such an approach, it has been possible to extend the mathematics used to both conventional and any other conceivable types of electrical machines.
In addition to electromagnetic processes, consideration
is given to the thermal, aerodynamic, hydraulic and mechanical processes associated with electromechanical and
mechanoelectric energy conversion.
In view of the importance attached to the above accompanying processes, the text discusses general aspects of
machine design and engineering.
The chapters on transients are based on the theory of
a generalized machine. The material includes the derivation
of differential equations for induction and synchronous
machines in terms of the d, q, 0 and the el, B, 0 axes, and
their transformation to a form convenient for computerassisted analysis and design.
The chapters dealing with specific types of machine
(induction, synchronous, d.c.) are largely concerned with
the conventional design. In each case, however, there is
a short discourse on the operating principle and arrangement
of the most commonly used special-purpose modifications.
The electromagnetic processes occurring in conventional
a.c. machines are described in terms of the resultant complex functions of electric-circuit parameters or their projections on the axes of a complex plane. As far as pr acticable,
il unified or generalized approach has been taken to deve-



loping equations for and describing the physical processes

in the two basic types of machine-the induction machine
and the synchronous machine . This concerns electromagnetic torque, electromagnetic active and reactive power ,
saturable magnetic circuits, machine inductances, etc.
More space is given to thyristor-controlled machines
gaining an ever-wider ground, than to a .c. collector machines
used on a limited scale .
In the ligh t of new findings, the effect of core saliency
on the harmonics of the airgap flux density has been treated
in a more rigorous form . A nove l approach has been taken
towards the equations for mmfs, emfs , electromagnetic
forces, electromagnetic torque, and machine characteristics .
Among other things, the equations for the synchronous
salient-pole machines are developed in terms of the d, q, 0
axes, the analysis of transients includes the short-circuit
condition in the synchronous generator, the starting of the
induction motor, and events in the single-phase motor .
The material marked with an asterisk (*) may be omitted
on first read ing, without disrupting the integrity of the
A. V. Ivanov-Smolensky

Introd uction


Basic Definitions

The utilization of natural resources inevitably involves

the conversion of energy from one form to another. Quite
aptly, devices doing this job by performing some mechanical
motion may be called energy converting machines. For
example, heat engines convert the heat supplied by the
combustion of a fuel into mechanical energy.
In fact, the same name goes for devices converting energy
in one form into energy of the same form but differing in
some parameters. An example is a hydraulic machine which
converts the mechanical energy of a reciprocating fluid
flow into mechanical energy further transmitted by a rotating
A sizeable proportion of the energy stored by nature in
chemical compounds, the atoms and nuclei of substances, the
flow of rivers, the tides of seas, the wind, and solar radiation
is now being converted to electric energy. This form of
conversion is attractive because electricity can in many
cases be transmitted over long distances, distributed among
consumers and converted back to mechanical, thermal, or
chemical energy with minimal losses. However, at present
thermal, chemical or nuclear energy is converted directly
to electricity on a very limited scale, because this still
involves heavy capital investments and is wasteful of power.
Rather, any form of energy is first converted to mechanical
by heat or water machines and then to electricity. The final
step in this sequence-conversion of mechanical energy to
electricity or back-is done by electrical machines.
From other electromechanical energy converting devices,
electrical machines differ in that, with a few exceptions,
they convert energy in one direction only and continuously.
An electrical machine converting mechanical energy to
electricity is called a generator. An electrical machine
performing the reverse conversion is called a motor. In
fact, a generator can be made to operate as a motor, and
a motor as a generator-they are reversible. If we apply
mechanical energy to the movable member of an electrical



machine, it will operate as a generator; if we apply electricity, the movable member of the machine will perform
mechanical work.
Basically, an electrical machine is an electromagnetic
system consisting of a magnetic circuit and an electric
circuit coupled with each other. The magnetic circuit is
made up of a stationary and a rotating magnetic member
and a nonmagnetic air gap to separate the two members.
The electric circuit can be in the form of one or several
windings which are arragned to move relative to each other
together with the magnetic members carrying them.
For their operation, electrical machines depend on electromagnetic induction and utilize the electromotive forces
(emfs) that are induced by periodic variations in the magnetic field as the windings or magnetic members are rotated.
For this reason, electrical machines may be called electromagnetic. This also applies to devices that convert electric
energy at one value of current, voltage and/or frequency
to electric energy at some other value of current, voltage
and/or frequency. The simplest and most commonly used
electromagnetic energy conversion device which converts
alternating current at one voltage to alternating current
at some other voltage is the transformer. Its coils and core
remain stationary relative to each other, and periodic variations in the magnetic field essential for an emf to be induced
in the coils are produced electrically rather than mechanically.
Electromagnetic energy converting devices with moving
or, rather, rotating parts are more customarily called rotary
converters. They do not differ from electrical machines in
either design or the principle of operation. In fact, rotary
converters can sometimes double as electric-to-mechanical
(or mechanical-to-electric) energy converting machines.
Therefore, we may extend the term "machine" to transformers
and rotary converters as special kinds of electrical machine.
Apart from electromagnetic electrical machines, some
special applications involve the use of electrostatic machines
in which the electromechanical conversion of energy is based
on electrostatic induction and utilizes periodic variations
in the electric field of a capacitor in which the plates are
free to move relative to one another. However, electrostatic
machines are no match for electromagnetic machines in
terms of size, weight and cost, and are not used in commercial
' or industrial applications.


1-2 Conversion of Energy by Tra nsformer

As energy converters, electric al m achines are import an t

elements in any power-generating, power-consuming, or
in dustrial installation . They are widely used as generators"
mot ors, or rotary converters at electric power stations,
factories, farms, railways, automobiles , and aircraft. They
are finding an ever increasing use in automatic control
Electrical machines are clas sed into alternating-current
(a.c.) and direct-current (d .c.) machines, according as they
operat e into or from an a.c. or a d.c. supply line.

Conversion of Electric Energy

by the Transformer

In sket ch form , the arrangeme nt of a simple single-phase

two-winding transforme r is shown in Fig . 1-'1. As is seen ,
it consists of two windings , land 2, wi th turns WI and W 2 ,


F ig. I-I Electric and magnetic cir cuits of a tra nsformer

which are wound on a magnetic core . For better coupling

between t he coils, the core is assembled from laminat ions
punched in electric-sheet steel having a high relative permeability, !-tT' with no air gap left around the magnetic
circuit. The laminations or punchings are made th in in
order to reduce the effect of edd y currents on the ma gnetic
field which altern ates at an angular frequency co , Let us
open , say, coil 2 and connect coill t o a source of a sinusoidal
alternating current of frequency f = Ul/2n and of voltage
VI = V 2 VI cos Ult, where VI is the rms value of voltage.
This will give rise to an alternating current, i1 = io, in the
coil , which can be found from the voltage equation for the '



circuit where it is flowing:


where R 1

- 81

+ R io


= resistance of winding 1

8 1 = -d'P'11/dt = emf of self-induction

1V11 = w 1c!) = flux linkage
cD = BA c = magnetic flux
B = magnetic induction (magnetic flux density)
A c = cross-sectional area of the core.
On setting flr constant and applying Ampere's circuital
law to the magnetic circuit

~ HI


(B/flrflo) dl = cD/AIJ. =



where Aj.1 = ~lrfloAc/lc is the permeance of the core and

lc is the mean core length, it is an easy matter to find the
inductance of winding 1
L ll = w1 CP/i o = w:Aj.1
" nd t he mutual inductance
L 12

wlIJ/i o


and to express in their terms the flux linkage

and the emf
81 =

-L 11 dio/dt

Using Eq. (I-i) and neglecting R 1 i o, we obtain the magnetizing current

io =

V2 locos (rot '-- n/2)

which produces an alternating magnetic flux

cD = i Ow 1 A j.1
Variations in the flux cD linking coil 2 induce in the latter
a sinusoidal emf of mutual induction

= -d1V 21/dt = -L 12 dio/dt

Thus, coil 2 can be used as a source of an alternating current

of the same frequency t. but at another voltage, V 2 = 8 2,
As is seen, the ratio of the instantaneous and the rms
emfs across windings 1 and 2 and of the respective rms vol-


1-2 Conversion of Energy by Transformer

tages , is equal to the turns, or transformation, ratio:

el/e 2 = E 1/E 2 = V 1/V2 = WI/W 2
If VI is specified in advance, we may use Eq. (1-3) to
find the turns numbers WI and W 2 such that V 2 will always
have the desired value. Winding 2 can be used as an a .c.
source by connecting it across a load resistance, R L . Then
the emf e 2 will induce in it a sinusoidal alternating current
i 2 = e2/(R L + R 2 )
which can be found from the voltage equation for the circuit
thus formed *
where V 2 = R L i 2
The secondary current i 2 will bring about a proportionate
change in t he primary current i l . The relationship between
i l and i 2 can be established by again using Ampere's circuital law written by analogy with Eq. (1-2) and recalling
that in a transformer under load both windings contribute
to the magnetic flux

~ Hz dl = (J)/A)J, =

i 1wI

+ izw z


Also, in writing the voltage equation for the circuit containing coil 1
and neglecting R1i l as in Eq. (I-i), we find that under load
the emf e l remains about the same as when coil 2 is opencircuited. This implies that el is induced by variations in
the same flux (J) and in the same magnetizing current i o
in coil 1, as exi st when coil 2 is open-circuited. If so, we
may equa te the right-hand sides of Eqs. (J-f ) and (1-5)
and argue that the sum of the magnetomotive forces in
coils 1 and 2 is equal to the mmf due to the magnetizing
current i o in coil 1
In an adequately loaded transformer with a closed (noairgap) core, iOw l is negligible

I iOwl I ~ I i1w1 I ~ I i 2w2 I

* Here and in Eq. (1-6), the emfs induced by leakage fluxes are
not included.
2- 0 16 9



So, without introducing an appreciable error, we may set


= 0

On this assumption, the directions of currents in the

windings are such that their mmfs balance each other:
It follows from Eq. (1-8) that the ratio of the absolute
values, [ r ], and of the rms values, I, of the currents in
coils land 2 are inversely proportional to their turns ratio
Using Eqs. (1-3), (1-4), (1-6), and (1-8) and neglecting
the losses associated with the cyclic magnetization of the
core and with variations in the energy of the magnetic
field, let us consider the balance of the instantaneous powers
in the transformer. The power delivered to coil 1 by the
supply line is
PI = vIiI = -eli l

Some part of this power, iiRI, is dissipated as heat in coill,

and the remainder, -eli l = ezi z, is transferred by the
electromagnetic field into coil 2. The power supplied to
coil 2

is partly dissipated as heat (i~R2)' whereas the remainder,

vzi z, is delivered to the load.

Electromechanical Energy Conversion

by an Electrical Machine

In sketch form, the arrangement of a simple rotating

electrical machine is shown in Fig. 1-2. As is seen, it consists of a stationary member called the stator, and a rotating
member called the rotor. The stator core, 4, is made fast
to a base-plate, whereas the rotor core, 3, is mounted on
a shaft carried in bearings, so that it is free to rotate, remaining aligned with the axis of the stator. On its cylindrical
surface, the rotor core 3 has slots which receive a single-coil
rotor winding, L, with turns WI' The stator core has similar
slots which receive a single-coil stator winding with turns W 2.


1-3 Energy Convers ion by Electri cal Machine

The stator and rotor cores are assemb led from ringshaped laminat ions .py nched in electrical-s.heet ste~l hav ing a high parmeahil it y for better magnetic coupling between the windings. For the same purpose, t he coils are
sunk in slots rather than put on the outer surface of the

Fig. 1-2 Electric and magnetic! circuits! of a simple electrical rria~
chin e Irr[the generating mode (il > 0, T em < 0) . . .

cores . With this arrangement, the air ga p between the stator

and rotor may be made very small and the magnetic circuit
presents a very low reluctance .
The shaft carrying the rotor coup les 'it to another machine
with which it exchanges mechanical energy (deliveri ng it
in motoring, and receiving it in operation as a generator).
The stator and rotor windings are connected to lines with
voltages V 2 and VI' respectively. In motoring the li nes (or
one of them) deliver electric energy to the machine. In
operation as a gene rator, the machine delivers electric
energy to the li nes (or one of them ).
Electromechanica l energy conversion by an electr ica l
machine utilizes the emfs t hat are induced in the windin gs
as a result of variations in their relative position in space.
To begi n with, su ppose t hat wind ing 2 is energized with
i 2 = constant, and winding 1 is open-circuite d, so t hat i l = O.
In t he circumstances, a stationary magnetic fiel d is set up ,
with its north pole, N , l ocat ed in the bottom part , an d the
south pole, S, in the top part of the st ator core,



Assuming that the permeability of the stator and rotor

cores, !-La,c is infinitely large in comparison with that of
t he air gap, !-Lo (!-La,c ~ !-Lo) , we may neglect the magnetic
potential difference across the core. Then, on writing
Ampere's circuital law for any loop enclosing the current
i 1w 2 in coil 2 (for example, the loop shown by the dashed
line in Fig . 1-2)I

we find the magn etic induction in the air gap due to coil 2
to be
B 2 = !-LOi2W2/28


where 5 is the radial air gap length.

The flux linkage 112 of this field with winding 1 varie s
with the angle y that it m akes with winding 2. When y = 0,
the flux linkage h as a m aximum positive value
lf12. m

B 21:lw 1


where l is the core length in the axial direction and 1: = scR

is the pole pi tch.
As the ro tor turns through an angle y anywhere from zero
to 180, the flux linkage varies linearly as a function of the
angle y
lf 12 = lf 12 , m (1 - 2y/n)
Wh en y = n/2 , t he flux linkage is zero, lf12 = O. When
y = n, it has a m aximum nega tive value , lf 12 = -lf 12. m
As the rotor keeps rotating, the flux linkage builds up
linearly as a function of the angle y
lf 12

-lf12, m (3 -2y/n)


and completes a period of variations when y = 2n.

The mutual inductance between the win dings, L 12 = lf12/ i 2,
varies in a similar way:

L 12
L 12

= L 1 2 , m (1 - 2y/n)

-L 1 2 ,m (3 -


for 0
for n

< Y<
< y<



where L 12 m = !-LOW1W2h/25 is the maximum mutual inductance between the windings.

1-3 Ene rgy Conversion by Electrical Machine


If the rotor is turning with an angular frequency Q, the

angle I' = Qt increases linearly, so that the emf induced
in winding 1 is given by
e1 = -d'12/dt = - i 2 dL l 2/dt = - i 2Q dL l2/dl' (1-15)
It is called the emf of rotation or the motional emf.
As is seen, the motional emf is proportional to the angular
displacement, angular frequency and derivative of the
mutual inductance with respect to the angular displacement
of the rotor. From Eqs. (1-14) and (1-15) it follows that
e1 = (2/:rr.) L12 ,mi2Q
0 < I' < :rr.
e1 = -(2/:rr.) L12 ,mi2Q
rr < I' < 2:rr.
sign applies when the emf is in the positive direction
of the current in coil 1; the "-" sign applies when it is in
the negative direction. The positive directions of currents
in windings 1 and 2 are such that the magnetic fields are
directed upwards as in Fig. 1-2, with I' = O.
Thus, with i 2 held constant, a square emf waveform is
induced in winding 1 of the elementary machine. The flux
linkage, mutual inductance and emf vary with a period
T = 2:rr./Q. Hence, these quantities vary with a frequency
given by
f = Q/2:rr.


Using Eqs. (1-10) and (1-14), we can express t he motional

emf defined by Eq. (1-15) in terms of the magnetic induction B 2 in the air gap
e1 = 2B 2luw1
0 < I' <:rr.
where U = rQ is the tangent velocity at the middle of the
air gap. Therefore, the direction of e1 can be determined
not only from Eq. (1-15). using Lenz 's rule, but also using
the right-hand rule. Of course, both approaches give t he
same result (Fig. 1-2).
If we, now , conn ect winding 1 having an internal resistance
R 1 across' a load resistance R L , the circuit thus formed
will carry an alternating current given by
t, = e1/(R L
R 1)
varying with the same frequency f as e1 does. The power
generated in winding 1 will then be
P,li 1 = - i ,i 2Q (dL I 2/dl') = (VI
i 1R 1) i]



Some of this power, iiR I , will be dissipated as heat in

winding 1; the remainder
PI =


= iiRL

will be delivered t o load.

The voltage across winding 1,

ilR L

which is the same as the load voltage, will likewise vary

with frequency f.
On the assumption that i 2 is constant, winding 2 is energized from a source of d.c . voltage
The power

it receives does not undergo electromechanical conversion

and is comp letely dissipated as heat .
The interaction of the magnetic fields set up by i 2 and i l
produces an electromagnetic torque T em acting on the
rotor. In determining T em' we may proceed from the fact
that the work it performs as the rotor is turned through
a sma ll ang le dl' is equal to the change in the energy of the '
magnetic fie ld, dW, caused by a change in the mutual inductance , dL 1 2 , assuming that both t, and t remain constant, or mathematically
T ern

dl' = dW = i Ii 2 dL I 2

H ence,
Tern =

i Ii 2 dL 12/dl'


If the angular dis placement of the rotor, d'l" is in the

direction of rotation, the torque in Eq . (1-19) acts likewise
in the direction of rotation and is posit ive. If dl' is in the
opposite direction, the torque in Eq . (1-19) is in the opposite
direction, too, and negative. In the generator mode of operation , the torque is, as is shown in Fig . 1-2, negative, T ern < O.
Using Eqs . (1-10) and (1-14), we can express the electromagnetic torque in terms of the magnetic induction in the
air gap , B 2 , as well:

(I -20)

1-3 Energy Conversion by Electrical Machine


The direction of the tangential electromagnetic force,

= 2B zli l w l , and of the torque in Eq. (1-20) can be ascertained, using the left-hand rule, as is done in Fig. 1-2 for
the generator mode.
Under steady-state conditions, when the rotor is spinning
at a constant frequency Q, the electromagnetic torque Tern
must be balanced by an external (mechanical or load)
torque T m
T m = -T em = - i l i z (dLlz/dy)
For this to happen, a mechanical power must be applied
to the rotor via its shaft
T m Q = -ilizQ (dL l2/dl')
which is converted to an equal electric power, eli l, given
by Eq. (1-18).
When y is anywhere between zero and 180, both i z and i l
are positive and dL 12/dy < O. In contrast, when y is anywhere between 180 and 360, i z is positive, i l is negative,
and dL 12/dy> O. Accordingly, the power in Eq. (1-22)
is positive, T mQ > 0, not only when the rotor takes up the
position shown in Fig. 1-2, but in any other angular position . This implies that an elementary electrical machine
can perform electromechanical conversion of energy in one
direction only (in our case, it can only operate as a generator).
The same elementary machine can operate as a motor,
thereby converting electricity to mechanical energy. To
this end, winding 1 must be connected to an a.c. supply
line of voltage VI and frequency [, so that i l is always in
opposition to el (Fig. 1-3). On writing the voltage equation
for the circuit thus formed


-e l

+ ilR I

and multiplying it by i. . we obtain the power delivered by

the supply line to winding 1:
VIiI = -eli l

+ i~Rl

Some of this power, iiRI' is dissipated as heat in winding I

and the remainder
-eli l = ilizQ (dLlz/dt)
is converted to mechanical power

T em Q = ilizQ (dLl2/dt)



transmitted by the rotor to the shaft of the driven machine.

Using the right- and left-hand rules, it is an easy matter to
see that in motoring the torque is positive (T em > 0) and
is in the direction of rotation.

Fig. 1-3 Electric and magnetic circuit s of a simple electrical ma chine

in the motoring mode (il < 0, Tern> 0)

To sum up, the elementary electrical machine we have

examined is reversible-it can operate as both a generator
and a motor. This is in fact true of any electrical machine.

Functional Classification
of Electromagnetic Energy
Converting Devices

The analysis of simple electromagnetic energy converting devices set forth in Sees . 1-2 and 1-3 shows that transformers and elementary electrical machines can only operate
from an a.c. supply line operating at frequency f.
If a transformer or an electrical machine is to convert d.c.
'energy , the d.c. supply must first be converted to an a.c.
form by a suitable device. This may be a semiconductor
device, or a mechanical one as in electrical machines (in
the form of a commutator whose bars are connected to the
respective coils of the rotating winding, and fixed brushes
riding the commutator).


1-4 Classification of Electrical Machines

Tab le 1-1 Functional Classification
of Electromagnetic Energy Conve rting Devices
BIocI( d ia gram



D. c.-e-n. C .

Conversion of al terna ting curren tat one voltage to alternating current at another voltage



A. C.-to-D.C.
D. C.-to-A. C.


Function(s) performed

Conversion of
d.c., or back



A. C. el ect ri cal

---- ---

D. C. electrical
machine (commutator- or rectifier-type)

---- --

A. C. rotary
(A. C.-D. C .
electrical ma chine)

A. C.-to-D . C.
rotary con vertel'

'V f


CD f

o-;--D> ,Q


a.c. to

Conversion of d .c . at one
voltage to d.c . at an other voltage

m echanical

of a .c, to
energy (or


of d ,c . to
ener gy (or

Conversion of a.c, at 11
to a.c . at 12 =f= / 1 and to
mechanical ene rgy (or
in an y other diroc tion)

Conversion of a.c , at II
to d.c. or mechanical
energy (or in an y other
d irec ti on)


Table 1-1 (cont inued)

Descr i p t io n

B l ock d i agr am

D. C. rotary
o--I--e:t:> 52

F un ction(s ) p erfor m ed

Conversion of d.c, at V~
to d.c . at V2 =FV l and
to mechanical energy
(or in any other direction)

If we consider an electromagnetic energy converting

devi ce in combination with a rectifier as an entity performing
a specif ic funct ion, we shall obtain a func tional classification as given in Table 1-1.


A n O utline of Tr a nsformers

Purpose, Appl ications, Ratings

A transformer is an electromagnetic energy converting device which has no moving parts and two (or more) windings
fixed relative to each ot her, int end ed to tra nsfer electric
energy between circuits or systems by virtue of electromagnetic induction .
El ect ric energy in the form of an alternating current
taken from a supply line with m l phases at a phase voltage
V l and frequency t, is impressed on the input , or pr imary,
winding whence it is transferred by a magnetic field into
the output, or secondary , winding with m 2 phases at voltage V 2 and frequency f 2. In most cases , transformers only
change voltages, VI =1= V 2 ' or currents, II =1= 1 2 , without
affecting freq uency or numb er of phases.
As a ru le , there is no conductive connection between
the primary and secondary windings, and energy transfer
between t hem is only by ind uc t ion ("transformer action").
A transformer having two single- or polyphase windings
with no conductive connection between them is termed a
two-winding transfo rmer (Figs. l -la and 1-2, respectively).
A t ransformer having three or more win ding s (Fig. l -lb )
with no conductive connection between them is called a
threeioinding transformer or a mul.tuoinding transformer
(see Sec . 10-1) .
Standing apart from other transformers is the autotransformer in which some of the energy delivered by a supply
li ne is transferre d Lo t he secondary winding conducti vely
(see Sec. 10-2) owing to a connection between the primary
and secondary sides .


Part One. Transformers



r\ ---,

.....- r-:-

..s-- ' /







Fig. 1-1 Single-phase transform ers: (a) two-winding and (b) threewinding:
I - primar y winding; 2-secondary winding; 3-secondary winding ;
4-m agneLic circuit (core)

Fig . 1-2 Three-phase, two-winding transformer :

I-star-connected three-phase primary coil s; 2-star-connected threephase secondary coils; 3-magneti c circuit (core)


db. 1 An Outline of Transformers

As already noted, energy supplied by a line is impressed

on the primary wi1!-ding which m.ay be sin.gle-.or polyphas~.
If energy converSIOn proceeds in the direction shown in
Figs. 1-1 and 1-2, the primary windings are those which are
labelled by the numeral "1". The secondary windings deliver
power to a load line; in Fig. 1-1, windings 2 and 3 are the
secondary windings, and in Fig. 1-2 it is winding 2. As is
seen, a multiwin~in~ transformer may have several prima~y
and secondary windings, For example, the transformer in
Fig. 1-1b has two secondary windings, 2 and 3.
polyphase windings are formed by star- or delta-connecting the phase windings of which there are as many as are
phases in the supply line. Each phase winding is a multiturn
coil mounted on a separate limb (or leg) of the transformer core.
In terms of phases, there may be single-phase transformers
(Fig. 1-1a and b), three-phase transformers (Fig. 1-2), and
polyphase transformers.
As electric energy converters, transformers have found
many uses. Among other things, they are involved in the
transmission of power from electric stations to consumers.
As often as not, this calls for the voltage to be stepped down
or up more than once. Therefore, the overall installed capacity of transformers in present-day electric systems is five
to seven times the installed capacity of generators.
Apart from transformers and autotransformers used in
power transmission and distribution syst ems and referred
to as power transformers, wide use is made of transformers
intended to transform the number of phases and frequency.
Also, special-purpose transformers are used in various industrial installations, communications, radio, television,
aut omat ic control, and measurements.
Commercially available transformers are made with power
ratings from fractions of a volt-ampere to several hundred
megavolt-amperes, for voltages from fractions of a volt to
several hundred kilovolts, for currents up to tens of kiloamperes, and for frequencies up to several thousand hertz.
Among special-purpose transformers are pulse transformers,
variable-voltage transformers, stabilized-voltage transformers, etc. (see Chap. 15).
Transformers are manufactured to relevant specifications
or standards and are designed to perform specific functions.
Accordingly , they are rated in terms of frequency, current,


Part One. Transformers

vollage, power, or some other values, all of which are called

ratings or rated values. They are given on a nameplate attached to each transformer . In this text we shall denote them
by the subscript "R" .
The voltage rating, or rated voltage, is the line (or lineto -line) voltage as measured across the line terminals of
a particular winding, and is designated (in Soviet practice)
as Vt, R, line or V 2. R, line.
The power rating, or rated power, of a transformer is
its total power, which is

8i, R

= V 1, R

l i,R

for a single-phase transformer, and

8 i, R

V3 Vi, R. line]i, R,


= 3V i , Rlt, R

for a three-phase transformer* .

In a two -winding transformer, the power rating, or rated
power, of the primary winding, 8 i , R, is the same as that of
the secondary winding, 8 2 , R, and equal to the power rating
of the transformer, 81, R = 8 2 , R = 8 .
The rated frequency, JR' of a harmonically varying quantity (current or voltage) for general-purpose transformers
is 50 Hz in the USSR and 60 Hz in the USA and some other
countries .
Rated currents are found from the power rating and the
rated voltage of the respective winding:
I i R =

8 RIVI , R

for single-phase transformers,

Ii. B , line = 8 R I V3 Vi. B , line

for three-phase transformers (line current) and

I~ :


8 R /3V i , R

for three-phase transformers (phase current).

The nameplate data are not to be understood as a prescription to operate the transformer only at its rated capability . Actually, its secondary current is allowed to vary

* Here and elsewhere in the text, the line quantities have the
subscript "line", whereas the phase quantities have no subscript. For
example, VI, line is the primary line voltage and VI is the primary
phase voltage .

Ch.1 An Outline of Transformers


from zero to 1 Z, R , with short-duration overcurrents [13J.

Also, app licable standards permit sli ght variations in
voltage and frequency .
It is to be noted t hat if we hold the primary vol tage constant, the secondary voltage will vary with the magnitude
and nature of load and may differ from its value at no-load
(open-circuit voltage) , when the secondary current is zero.
It would seem that the rated secondary voltage should be
taken equal to t hat at the rated power SR' Unfortunately,
this voltage depends on the phase relation between the
secondary current and voltage . Therefor e, to avoid ambiguity, the rated secondary voltage , Vz,R, is taken to be
equal to the no-load (open-circuit) voltage, when the secondary current is zero.
Arbitrarily , the rated secondary current is t aken to be
equal to that computed from the rated power at the rated
secondary voltage :
1 2, R = S RIV2 ,R
for single-phase transformers,

1 2 R , line = SRI V3V 2 , R . line

for the line cur rent of a three-phase transformer, and
1 2 ,R = S R /3V 2,R
for the phase current of a thr ee-phase transformer.
A transform er can step up or down the applied voltage.
In a step-up transformer, the primary wind ing is on the lowvoltage (LV) side, an d the secondary winding is on the highvoltage (HV) side . In a step-down transformer, they are arranged the other way around. For ex ample, the transformer
in Fig. 1-2 will be a step-up one if Vl ,R is lower than V 2,R,
or a step-down one if V l, R is higher than V 2,R (the arrows
in the figure show the direction of powe r tra nsfer ).

Construction of a Transformer

(i) The Core and Coils

The actual energy conversion in a t ra nsformer t akes place

in its core and coils .
For better energy conversion , the coils are placed on,
or enclosed in , a magnetic circuit fabricated from a ferromagnetic materia l having a high permeability , [La' which



Part One. Transformers




Fig. 1-3 Transform er windings:

(a) coaxia l and (b) interleaved


Fig. 1-4 Two-l ayer cylindrical

strip-conductor winding

is hundreds of times that of

~ free space, ~~o (see Figs. 1-1

and 1-2). To have a high permeability, the magnetic circuit ought not to be excessively saturated , and its magnetic induction (magnetic flux
density) at a maximum magnetic flux ought not to exceed
'1.4 to '1.6 T* . The required
reactive power can be reduced
by minimizing the leakage
!luxes each of which links
with only the primary or only
the secondary winding. One
way to reduce leakage fluxes
is t o reduce the gap between
the primary and secondary
windings. To this end, the
primary and secondary coils
of a phase are put on the same
leg or limb (see Figs. 1-1 and
1-2). The windings may be in
the form of cylindrical coils
taking up the whole length
of, and arranged coaxially
on, the limb (Fig . 'i-3a) or
as a series of pancake or disc
coils with the primary and
secondary sections alternating
in an interleaved or sandwich
arrangement (Fig. 1~3b). Of
a larg.e number of various
coaxial windings, the cylindric al winding is the simplest
(Fig . 1~4).
An important aspect in improving the efficiency of energy

* T stands for the tesla, tho

unit of magnetic flux density in
the International Sys tem (51).Tran slator' s note,


Ch. 1 An Outline of Transformers

onversion is to reduce the amount of power lost as heat.

~o this en d, t he win di ngs are m a.d e of a m a terial \~i th a 10:,"

resistance an d a large cross-sect ional ar ea , an d with a n11nimum acce p table turn length.
The . m agn etic circ uit is designed so as to kee p eddycurrent and hyst eresis losses to a minimum . Th is is usually
done by using magnetically soft , ele ctrical-sheet steels wh ich


(b )

Fig. 1-5 Si ngle-phase transfo rme rs : (a) core typ e and (b) core-andshell (five-leg cor e) typ e:
i -limb (leg); 2-yoke; 3-outer li mb (leg)

have a low hyst eresis l oss and hi gh res istivity, an d assem bling the core fr om indivi du ally insul ated l am inati ons with
a thicknes s ch osen such tha t eddy currents would not affect
the mai n magne tic fie ld an d woul d not lead to increased
eddy-current loss. The lam ination thick ness d depends on
the frequency t of t he magnetizing current (see Sec . 31-3) ,
and is t aken as 0.35 mm 0 1' 0.5 mm for 50 Hz .
With a core fabricated as outlined above , t he iron (or
core) loss can be kep t at a level comparable with the copper
loss, an d the dem agnetizing effect of edd y cur rents can be
reduced t o a mi ni mum .
Transformer cores mos tly come in anyone of two designs, the core type an d t he shell type. In a core-type single phase transformer, the core cons ists of two vertical li mbs
around wh ich the prefo rmed circular windings are pl aced.
The win dings cons ist each of two coils which m ay be connected in serie s or par allel an d are pl aced on different limbs .
The .top an d bott om memb ers , called the yokes , join the
two li mbs in to a closed magnet ic circuit (see Fi g . 1-5a).
3- 0169


P art One. Transformers

In a core-type three -phase transformer, a primary and

a secondary winding of one phase are wound on each lim b
(see Fig. 1-2). The three equal limbs are join ed by the two
yokes into a closed magnetic path.
In a single-phase shell-type transformer , t he core is divided
so that parallel magnetic paths encircle the single group or
coils on two sides as if by a shell (see Fig. 1-1). As is seen,
the yokes ar e built up to a cross-sectional area half as large
as that of the limb carrying the coils .
To reduce height and to facilitate transit by rail, highp ower transformers have fiv e-leg core-type circuits calle d

Fig. 1-6 Three-ph ase core-and-shell -t ype transform er :

l-limb; 2-yoke; 3- oul er limb

the core-and-shell type in Soviet usage (Figs. 1-5b and 1-6).

A core-and-shell transformer is lower in height because the
yokes have to carry half as large a flux and may therefore
have a lower height, too. As an illustration, Fig. 1-5 shows
single-phase transformers of the core and the core-and-shell
type of construction having the same power rating. Th e
height can be reduced by about the same amount in a threephase core-and-shell transformer (Fig . 1-6) where the yok es
have to carry a flux which is 1/rS times that in the limbs.
In core-type transformers, the yok es carry the same flux
as the limbs do .
At t he corners of a core , t he yokes and limbs may be
. joined in any one of two manners. One gives butt jo ints;
and the other, interleaved (or imbricated) joints.
With butt joints, the limbs and the yokes ar e stacked up
sep arately, the coils ar e put on the lim bs , and the top yoke

Ch. 1 An Outline of Transformers


is then placed on (joined with) the limbs to form a closed

magnetic circuit. The butt joints are filled by insulating
spacers to avoid eddy currents at those places . The spacers
form a virtual air gap which absorbs reactive power over and
above that required by the iron itself. Because of this, butt
joints are seldom used, although they simplify assembly and
disassembly .
Interleaved (or imbricatedi joints are used on a wider scale.
In this case, the successive layers of laminations in the
yokes and limbs are interleaved so as to give an overlap at
the corners to reduce the joint reluctance (Fig . 1-7). Even so,
the flux has to cross the insulation between the laminations
at the overlapped portions, but the virtual air gap thus
formed absorbs less reactive power than a core of the buttjoint type. A disadvantage of the interleaved type is that
a core already assembled has to be disassembled (unbladed)
at the top yoke so as to let the coils be put on the limbs.
After that the top yoke is assembled (rebladed) again.
Modern electrical-sheet steels disp lay directional (anisotropic) properties produced by cold rolling so that
in the direction of rolling they have a reduced specific loss
and an increased permeability [131. However, there is an
increase in loss and a reduction in magnetic intensity at
the joints between the limbs and yokes, where the magnetic lines of force turn through 900 from the direction of
rolling . This drawback can to a marked degree be minimized by using mitred joints or overlaps as shown in
Fig. 1-8.
In low-power, low-voltage transformers, the coils may
be wound on rectangular formers and the limbs may be
given a rectangular cross-section. In high-power transformers, the coils are wound on a cylindrical mandrel, and the
limbs are given multistep cruciform cross section approaching the area of the circumscribing circle so that the area
within the coils has a more efficient iron-to-air ratio
(Fig. 1-9). The yokes usually have a rectangular or a cruciform section with a limited number of steps.
Clamping and packing arrangements for transformers
vary from size to size. In power transformers rated at under
1 MVA (per limb), this is done with wooden or plastic battens
and bars which fill the space between the limb and the insulating sleeve carrying the LV winding which is placed next
to the limb iron (Fig .1-9a).


Part One. Transformers







~ lamination~


2''' ~

Fig. 1-7 Imbricat ed (interl eav ed) joints in a magnetic core :

(a) sin gle -phase core-type transform er ; (b) three-phase core-type tran s-

forme r


Fig . 1-8 Mitred joints for a three-phase core-t ype tr ansformer using
cold-rolle d gra in-orie nte d steel sheet lamin at ions

Fig. 1-9 Yok e clamping:

(a) by wooden bat tens; (b) by steel s tuds (i-steel st ud; 2-insula ting tube; 3-pressboard wash er; 4-steel washer; 5-pressboard

Ch. '1 An Outline of Transformers


In high-power transformers, the limbs were at one t ime

clamped by steel studs insulated from the iron by syntheticresin-bonded paper cylinders (Fig. 1-9b), whereas the yokes
were clamped with similar studs extending through wooden
or steel yoke clamps (Fig. 1-tO) .
The more recent practice is to clamp together the laminations in transformer limbs and, often, yokes with circumie-

Fig. 1- 10 Transformer frame

rential bands usually made of glass fibre bonded with thermoset ti ng epoxy compounds . (Such bands can be seen on the
limbs in Fig. 1-tO, and on the limbs and yokes in Fig. 13.)
With epoxy-resin-bonded bands , one nee d not use clamping
studs or punch holes in the core laminations (such holes
reduce the reluctance of the core and add to no-load losses) .
The core and the yoke clamps along with the other parts
serving to h ol d the core and coils in place make up the frame
of a transformer (Fig . 1-10) .
Microtransformers rated for units to tens of volt-amperes
use far simpler core designs. As often as not, their cores are
assemb led with one-p iece punch ings as shown in Fig . 1-11a ,
or two-piece laminations (one piece being E-shaped, and
the other l-shaped) as shown in Fig. 1-11b.


Pa rt One. Tra nsformers

In the lamination shown in Fig. 'i-Ha, the middle li mb

is cut through, so it can be bent away during assembly,


- -

== =r---- -


Fig . 1-11 Core assembly for microtransformers:

(a) a-shaped laminations; (b) E- and I-sh aped lami nations

and coils can be put on it and inside the laminat ion. The
next lami nat ion is inserted from the opposite end of the

Fig . 1-12 Strip-wound transformer :

I-primary; 2-secondary; 3-core

coil. After assemb ly, the core is clam ped t ight by pressure
end plates and studs.
Another popular core design is that using long strips or
ribbons of transformer steel wound on a r ing -sha ped former,
and the coils in turn wound on the core by a suitable machine
(Fig . 'i-12) .

Structural Parts of a Transformer

The fu nction of the structural parts in any transformer is

to provide electrical insulation between the windings,
to hold the core and coils in place, to cool the transformer,
to provide connection between the transformer windings
and the associated electric. li nes, and the like. Actually,
the yoke clamps .and the other clamping and packing-on t
parts may also be classed as structural parts.

Ch. 1 An Outline of Transformers


Let us take a closer look at !.he structural parts, choosing

all oil-immersed three-phase power tr ansformer as an example .
Its general arrangement is shown in Fig. 1-13.

Fig . 1-13 Three-ph ase, t wo-winding, 40-MVA, 110-kV transformer

with split LV windings and on-lo ad t ap-cha ngin g on the BV side:
I-HO-kV bu shing; 2- 1O-kV LV bushing; 3-lifting lu g; 4- tank;
5-tubular cooler ; 6-therm al siphon filter; 7- jacking lu g; 8- oil
dr ain cock; 9-blow er; IO- cas tors; ll-yokeb ands ; 12-frame tie-rod;
13-yok e cla mp ; 14 -HV ta p-ch anger ; 15-limb band s; 16- core-andcoil li ftin g lu g; 17 -conserva tor; 18-oil gauge ; 19 - expl osion stack

Wind ing insulation. The turns of a transformer winding

must reliably be insulated from orie another, from ' the
turns of other windings, and from the transformer frame .
In oil -immersed transformers for '10 kV and high er , this
purpose is served by oil-paper barrier insulation . It is
obtained by impregnating cab le pa per or electric-grade
pressbo ard with transformer oil whi ch is also used to fill

P art One. Transformers


Fig. 1-14 Windings of th e transformer in F ig. t -t3 (dimensions

in mm) :
1 - steel pressure ri ng; 2-fine
vo 1tage-control winding; 3-coarse
voltage-control winding; 4- H V
windi ng; 5- LV win ding; 6-corner
washer; 7-intercoil spacers; 8-insula t ing cyli nder ; 9-pressb oard
support rings ; 10 - pr essb oard yoke
ins ula tion; ll-edge bl ock ; 12wooden pac king strip; 13-wooden
fillin g b ars; 14 , 15-press-b oar d
cleat s

Yoke line


Y"o~l' un~J~~~~~!~~t





Radi l1L section of !f0ke


Ch, 1 An Outl in e of Transformers


the space bet ween t he coils an d the frame. Apa r t from pr oviding electrical insulation, the transformer oil fill ing the
transformer tank also doub les
as a coolant.
Int ertu rn in sulation is prov i7
ded by the oil-im pregnated in8
sulation on t he coil cond uctors
(which may be round wire or
str ip con ductors) .
The arrangement of the ma jor
insulation separating the windings fro m each other, from the
4tank , and from the frame is
shown in Fi g. 1-'14.
Leads and terminal bushings.
~.-rl..,-llIl vv<.l<"~10
The L.V . an d H . V . win dings of
a transformer are connected to
external circ uits by means of leads (insulated con ductors mounted ins ide the transformer tank)
and terminal bush ings (dev ices
cons isting of a porcela in cylinder, a central current-carrying
conductor, and a m ounting fl ange).
Fig . 1-15 Out door-service,
35 1,V, 250 A bushing wi th
The conductor of a t erminal
central conductor connected
bushing must reliably be insul atto a lead :
ed from the grounded top cover
l-copp er termin al ; 2-brass
on either ("oil" and "a ir" ) side
nut; 3-bruss cap ; 4-steel
st ud; 5-nut; 6-washer;
(Fig. 1-15). The si ze and com7- rubber gro mme t ; 8- por- plexity of terminal bushings grow
celain insulator;
flan ge; lO-lug; ll-rubbel' with the voltage rating of transformers . For '110 kV an d higher,
seal ; 12-central conductor
inside insu la tin g tu be
oil-filled terminal bushings are
Tan k accessori es an d fit tin gs . If the tank of an oil -immer sed transformer were filled full with oil and completely
sealed off, it woul d inevitably burst un der the action of oil
pressure bu ilding up in the tank with rising temperature.
One way to prevent bursting is to keep the oil level in the
tan k some distance below th e t op cover an d let the tank's
insid es communicate with the atmosphere . In such a case,
however 1 the oil is exposed to air over the entire area UlHIE;lr


Part One. Transformers

the cover-a feature which speeds up oil ageing through

oxidation and moisture pick-up, so the oil loses its valuable
properties too soon. Another course of action is to fit a
conservator (or expansion tank) to the tank-a cylindrical

Fig. 1-16 Accessories of a transformer tank:

1-oil gauge; 2-filler cap ; 3-breather; 4-sJudgc sump; 5-COllscrvator shut-off cock; 6-Buchholz relay; 7-relief stack

vessel communicating with the tank and limiting oil expo sure to air (Fig . t-:12). In a transformer with a conservator,
the oil needs to be dried, purified and regenerated or changed "far less "often .
The conservator is usually fitted "with an oil level gauge
(see Fig . 1-16), and a sump to collect sludge and moisture.
The space at the top of the conservator communicates with
the atmosphere via a breather tube brought out to terminate
under the conservator (so as to keep drops of moisture from
finding their way into the conservator).
Any transformer generates a large amount of heat in
operation, and this calls for a proper cooling arrangement.
On l arge transformers, this is done by tubular radiators
(see Fig. 1-13) which are attached to ports welded into the
tank. The ports are fitted with cocks so that the radiators
can be shut off and detached while keeping the tank filled.
The temperature of oil is indicated by a thermometer mounted in the top part of the tank. On small and medium-size

Ch. 2 Processes in Transformer at No-Load


transformers, mercury thermometers are used, whereas on

large units a better choice is filled-system thermometers or
remote-reading resistance thermometers with their indicators mounted at an instrument board.
Any fault which occurs inside a transformer in operation (insulation puncture, shorted turns, poor contact or
sparking due to poor grounding) is generally accompanied
by the evolution of gas as a result of the decomposition of
oil or solid insulation. The gas bubbles rise to the surface
and finally find their way to the conservator. On its way
there, the gas is collected in what is known as the Buchholz
or gas-formation relay (see Fig. 1-16) installed on a stub
pipe between the tank and conservator.
The Buchholz relay has an upper and a lower float. As
gas collects in the relay housing, it displaces oil out of it.
The top float drops, and its mercury switch completes an
alarm circuit . In the case of a more serious fault, such as an
inter turn short (or shorts) and the like, gas is usually liberated in an explosive fashion, and a large amount of oil is forced from the tank into the conservator. This causes the lower
float to rise and close its mercury switch, thereby activating
a tripping circuit which disconnects the transformer from
the supply line and averts a major breakdown.
To avoid irrepairable damage to the tank in the case
of a heavy gas evolution, a device known as the relief or
explosion stack is installed on transformers (see Figs. 1-13
and 1-'16). It is a long steel pipe communicating with the
tank at one end and closed by a disc of thin glass at the
other. When the pressure inside the tank rises dangerously,
the disc bursts, so that excess oil and gas are expelled into
the atmosphere 'before the tank has time to be deformed.

Electromagnetic Processes
in the Transformer at No-Load


The No-Load Condition

On the primary side, transformers are excit ed by a harmonically varying voltage



Part One. Transformers

As the load varies , the peak-value

F 1,m

and the frequency

I of the primary voltage ch ange but little , so it is usually

assumed that they are constant and equal to their rated
F1,m = V l ,mn = constant
I = In = cons ta nt
This also goes for t he angular frequency
W = 2'Jt1 = Wn = cons tant
The secondary current is inversely proportional to the
impedance of the line to which it is connec ted

I Z I = V H~+ X2
At a certain defin it e value of thi s impedan ce, Z
the second ary winding car ries its r at ed current
12 = 12 ,H
At Z


Zn ,

Zn, the secondary current exceeds its rated value



1 2 ,R

and th e transformer is somewhat overloaded .

At I Z I> I Zn l

1 2 < 1 2 ,n
and the transformer is underloaded . When I Z I is infinity,
which occur s when the transformer is disconnected from the
V, e, ~





F ig . 2-1 Single-p hase tw o-windin g transformer on no-l oad

receiving line on the secon dary side (the secondary is opencircuited), the secondary current falls to zero . In th e circumstances , the transformer supplies no-load current, which
is why this state is calle d the no-load (open-circuit) condition,

Ch. 2 Proc esses in Transformer at No-Load


The electromagnetic pro cesses occurring in a transformer

at no load ar e far simpler than they are und er load, with
10 > 0, so their study can best be begun with the no-load

Consider the electrom agnet ic proce sses at no-load in the

single-phase two-winding t r ansf ormer shown in sketch form
in Fig . 2-1. This is a core-t ype transformer whose primary
and secondary windings are shown for conv eni ence located
on different limbs. (The ac tual arrangement of the windings
on a core-type transformer has been described in Sec . '1-3,
see Fig. 'I-Sa.)

Voltage Equations

The supply voltage VI im pressed on the primary winding

gives rise in it to an alternating current i o , called the noload curr ent . This current produces two fluxes , namely the
mu tual (usefu l) magnetic f lux which has its path wholly within
the core of a very high permeabili ty, ~tr ~ '1, and links all
the turns WI an d W 2 of the primary and secondary windings,
and als o the leakage flux which links onl y the primary turns .
If we find the mutual magnetic flux cD at an y section of
the dosed magnetic circuit , we sha ll be able to find the
mutual flux linkage with the primary winding
1f on = w1 cD
and with the secondary winding
"If 02 1 = w2 cI)
The leakage flux has its path completed through nonmagnetic materials (air gaps, insulation) with a permeability equal
to that of free space , ~to, and substant ially smaller than that
of t he magnetic core . Therefore, the leakage flux linkage
with the primary winding at no-load , 1f a O, is a small fraction
of the mutual flux linkage with the primary, "If on (Fig . 2-'1 ).
Th e periodicall y varying mutual and leakage fluxes induce
electromotive forces in the windings with which they link.
For t he posi tive directions of currents, voltages, emfs and
magnetic lines of force shown in Fig . 2-'1, the primary emf
oj mutual induction is
e1 = - W I dCV/d t = - d"lf on /dt
whereas the secondary emj oj mutual induction is
ez = - w z dcD/dt = - d"lfOZl/dt


Part One. Transformers

and the leakage primary emf is

e ao = -elcDao/elt

~ e1


Interpreting VI as an emf impressed on the winding from

the supply line, we may write Kirchhoff's voltage equation

+ e + e ao =

R 1i o


where R 1 is the resistance of the primary winding.

The no-load voltage across the secondary is the same as
the emf induced in it
V 2 = e2


Variations in EMF with Time.

An EMF Equation

For all power transformers (and for most microtransformers) ,II

we may neglect in Eq. (2-5) both the voltage drop across R 1
and the leakage emf e cu

I u.i, I ~ I el I
I e ao I -e; I ell

and deem, with sufficient accuracy , that the primary emf

of mutual induction is in antiphase with the primary voltage (Fig. 2-2):


-VI =


V1 ,m cos rot
cos rot

Fig. 2-2 Time variations in

voltages, emfs and magnetic
flux of a transformer

It follows from Eq. (2-6)

that the emf of mutual induction . varies with time harmonically, and its peak (rms) value does not differ
from the peak (rms) value of the voltage

E 1 ,m = VI,m,

(E I = VI)


From a comparison of Eqs. (2-2) and (2-3), we may conclude that the ratio of e2 and l is time-invariant. This ratio
is called the transformation, or turns, ratio


= E 2 , mlEl,m = E 2/E1 =

W 2/Wl




Ch. 2 P rocesses in Transformer at No-Loa d

On the basis of E qs . (2-6) an d (2-8), we may arg ue that e2

varies likewise harmonically and is in phase with el .
We may express the magnet ic flux (D in terms of el by
integrating t he differential equation (2-2) subject to Eq . (2-6):

. ~
(1)= - - J e 1 d t=
\' cos or dz





- Q)msin ro t

(D m = E1 ,m/Wlro
is the peak va lue of the magnetic fl ux .
Using Eq . (2-10), we can der ive an equati on giv ing the
rms va lue of e1 from the given peak magnetic flux or flux
E 1 = E 1 n/ j/2 = row1cDm/ 1I 2 = ro'P'o11 ,m/ V 2
E 1 = (2n /

11 2) j W1(I)m

. (2-11)

Accordingly, the rms va lu e of e2 is

E 2 = rowlI)m/

11 2 = rolfo:d ,n,l 11 2


Referring to the plot of Fig. 2-2, the magnetic fl ux l ags

behind VI by 90 (it is said to be in qu adrature lagging with
the primary voltage) , and leads e1 and e 2 by 90 (it is said
to be in qua drature leading with the two emfs).

The Magnetization Curve

of the Transformer

The thick ness and m aterial of the lami nations for a transformer core are always chosen accord ing to t he frequency of
the magnetizing current , so as to keep eddy currents t o
a minimum . The instantaneous magnetic flu x may th en be
determined from t he inst ant aneous pr imary mm f, iOw 1,
at no-l oa d. The r esultant r el at ionship between t he in st ant aneous va lues of the two quantities, cD = j (i o), is i dent ical

Part One. Transform ers


to that obtained with d .c., when eddy currents are nonexi stent.
Graphically , the nonlinear re lationship betwee n t he flux
cD in the core an d the direct curren t i o in the primary winding is dep icted by what is called the cl.c. magnetization curve
(or characteristic) of a tran sformer. It can be construc ted on
the basis of Ampere's circuital la w in in t egral form . On
aligning the loop enc losing t he current in all the primar y
turns, iOw 1, with a line of for ce of t he mutual magnetic flux
in the core, Ampere's circuital law ma y be written

iowt =

~ Hzdl

The procedure yielding the cir culation of the H vector

is as follows:
(1) Assign a desired va lue to the magnet ic flux in th e
core .
(2) Bre ak up the core into n portions of length I" each ,
such that within each portion the act ive iron cross-sectional
are a A" and the permeabili ty r em ai n constan t for the specifi ed m agnetic flux .
(3) Calculate the magnetic induction within each port ion,
B" = <DIA" .
(4) Using the d.c . magnetization curve for each portion,
B = f (H), determine I h and ~L a " = B,JH h
(5) Adopt ~l a" = ~lo = 4:11 X 10-7 H im for the nonm agn etic gaps, and
(6) Replace the circulation ~ H, dl with the sum of
magnetic potential drop s across the individual portions,
spread over all t he ti portions:

iowt = :~ HI dl

"= 1


=:2; (Bh /~la") I"

"= 1

= cP :2;

I ,J~la"Ah =cD/Afl


" =1

where AJ.l

= 1/

~ Ih /~n hAh is the permeance of the core.

11= 1

On solving Eq . (2-13) for several values of cD and finding

each time i o = cDlw1A fl' we can then plot the magnetiza-

Ch. 2 Processes in Tranalor mer at No-Load


tion curve (J) = t (io) for the transform er. An approximate

sha pe of the m agnetization cu rve is shown in Fig. 2-3 which
also gives the A~t = t (i o) curve . As
is seen, A~ is a maximum near the
knee on the magnetization curve . As
i o and (I) keep increasing, the permeance decreases .

The No-Load Current


As already noted, if we neglect the

of eddy curren ts and the core
Fig . 2-3 Magnetizalosses, the rel ationship between the
t ion .cur ve, QJ = f (io),
instantaneous flux and no-load curof a transform er
rent is the same as with direct current. Therefore, since the magnetic flux has been found
to vary with time sinusoidally, as defined in Eq . (2-9),
we can use the d .c , magnetization curve shown on the left
of Fig. 2-4 in order to see how the no-load current varies


Fig . 2-4 No-lo ad cur rent wav eform (ignoring core Joss)

with time, i o = t (t) or t (wt) . To do this , we should plot

var iat ions in the magnetic flux with time, el) = ())m sin wt
(on the right of Fig. 2-4) an d determine the instantaneous
values of no-load curren t for some selected values of the
magnetic flux .
The relevant gr aphical procedure is indicated in the figure
by arrows . T aking the value of (J) at point 1 and moving
through point s 2, 3, and 4, we find the corresponding no-load
curre nt i o at the intersection, 5, of the horizontal extending
1, -0 1 GU


Part One. Transformers

from point 4 and the vertical extending from point 1 . With

a sinusoidal magnetic flux, the no-load current as depicted
by the i o = f (roz) curve turns out to be nousinusoirlal. The
no-load current grows increasingly more nonsinusoidal as
the peak magnetic flux cD m raises the level of saturation
in the core or, which is the same, as cD m exceeds increasingly
more the fluxes corresponding to the linear portion of the
magnetization curve where (I) is proportional to i o.

Transformer Equations
at No-load in Complex form

Complex notation is applicable to equations that connect

sinusoidal currents, voltages and emfs. Therefore, before we
may write the transformer equations at no-load in complex
form, we must replace the nonsi nusoidal no-Ioad current i o by a
sinusoidal current i or = V2 1 0 1'
)( sin wt (Fig. 2-5) equivalent in
terms of the reactive power consumed. (Since the copper loss in
the primary is low, the iron loss
is likewise low, so we may take
it that the active power and the
resistive current are small in
c;;z:....--4------=~-com parison with the reactive
power and the reactive current.)
For this change not to affect
the reactive power consumed, the
rrns value of the equivalent sinusoidal current, Ion must be
equal to the rms value of the
nonsinusoidal no-load current,
that is
Fig. 2-5 Replacement of
no-load current i o by an
equivalent sinusoidal current i Or

I 01' =

1 /

f Iiio d t



In our further discussion, we call this current the reactive

(magnetizing) component 0/ the no-load current . As is seen
from Fig. 2-5 , the current i or must be in quadrature lagging
with VI ' (A transformer at no-load and free from iron loss


Gil. 2 Pro cesses in T ra ns former at No-Load

Illay be rega rded with respect to the supply line as all inductor having a negligible ohmic res ista uco.)
As will be shown later, even if we in clude losses, the active
component of t he no-load current is very sm all in comparison with the reactive component , I Oa ~ l or . Therefore,
it is legitimate to deem the rms value of t he no-load curre nt,
1o, equal to t he rm s value of the re active current

10 =

-V I ij a + n; ~ 1

0 ,.

It i ~ important to note that the no-load primary current

usually ranges anywhere from 0.1 to 0.005 of the rated primary current .
For each rms value of the primary voltage , VI ' we find loT'
and compute the equivalent prim ar y inductance related
to the main or mutual magnetic flux or between the primary
and secondary. This will be referred to as the main or mutual
primary inductance . It is defined as the ratio of the peak
flux linkage to the peak reactive component of the no-load

If we express the magnetic flux in terms of the equivalent

perm ean ce A 1 2 of the t r ansformer


A 1 2 (w1 i or)

this permeance an d the mutual primar y inductance may be

connecte d by a relation of the form
From this inductance, we can com pute the mutual ituiucti oe
reactance of the primary winding

X 12

UlL~ 2 =

Ulw~ A 1 2


Now that we have introduced the necessary definitions,

we may write complex relations connecting the emf and
the sinusoidal reactive component of the no-load current .
As follows from Eqs . (2-2), (2-15), an d (2-16),


Part One. Transformers

N ow we shall express e1 and i or as tli e real parts of the

resp ec tive complex nnipl itud cs ruul tipli ed by ex p (j(f)t)*

[V 2" E1 exp (j(f)t )J

(2-1 8a)

ior = Re [V 2 i; exp (j(f)t) ]


e1 =


On substituting the above expressions in Eq . (2-18) and

differentiating, we obtain Eq. (2-18) re-wri tt en in complex
notation :
Re [V2E 1exp(j(f)t)] =


= Re


L~2 ~JV 2" j Or exp (j(f) t)]}

Re (-- j(f)L~2



ddt fRe

rv'2 i., exp (j(f)t )l)


E 1 = -j (f)L~ 2Ior = -jX 12Ior

Graphically, Eq . (2-19) and also Eqs . (2-6), (2-9), (2-11),
and (2-12), as wr itten in com plex notation

- V = E = - j(f)W d>n/ l! 2"

V2 = E 2= - j(f)w2(b~/V2

can be de picted by a phas or
diagram for a transformer at
no -load, such as shown in
Fig . 2-6 .
Fig . 2-6 Phasor diagram of a
transformer on no load (ignoring
losses and active current)


Mo-l oad losse s

Although the core of a transformer is assembled from thin

insulated electrical-sheet steel laminations, t he iron (core)
loss accounts for 0.1 % to 0.2 % of the transformer's power
rating. For example, in a 100-MVA transformer, the core
loss is up to 200 kW .
In microtransformers rated from 0.1 to 10 3 W, the core
loss rises to from 2 % t o 20 % of the power rating.

The symbols with a dot above refer to complex quantities.

CII. 2 Proc esses in Tr ansformer at No-Load


The core loss is"the sum of the hysteresis loss Ph which

is propor tional to" the frequency 1 and the square of the
magnetic induction , B~ , and t he edd y-current loss , P e which is proportional to the frequ ency squared , 12 , and the
magnetic induct ion also squared , B~ .
In practical calculations, it is customar y to find directly
the t ot al core loss, Peore, in t he various core elem ent s with
an active iron cross-sectional area A Ii , magnetic induction
B li = <Pm/Ali, and the iron mass m-,
P eore = Ph

+ e, = );

PI.olsoB~ (1 /50( 3 Inli

(2-21 )

where Pl.O/ 50 (in watts per kg) is t he spe cific core loss at
a frequency of 50 Hz and a magnetic induct ion of 1 T [13J.
At no-load and the rated primary voltage , VI R, the core
loss is about the same as at rated load. Th erefore (as will
be shown later), the magnetic flux and induction in the
core at VI are nearly in dependent of the lo ad condition .
The copper loss in the primary at no-load, PCu ,o = RJ~ ,
may be neglected , because the no-lo ad current is small and
t his loss is a fr action of that at the rated primary current
P Cu,l R . RIII,R

If we neglect the copper loss, the no-load loss of a transformer, Po, may be deemed equal to its core loss :

Po ~


The Effect of the Core Loss

on the Transformer's Perfo rmance
at No-Load

A transformer with the core loss P eore draws from the supply
line an active power given by TTI / oa' Th e nns value of the
sinu soidal active current is
lo a

P eore/TT I


This current is in ph ase with the applied volt age, so it may

altern at ive ly be expressed in te rms of an equivalen t resistance R l 2
lo a = F 11R 1'!.
From a com parison of Eqs. (2-22) and (2-23), the equivalent resistanc e R 12 m ay be expressed in te rm s of voltage


Pa rt One. Transformers

and core loss as follo ws*:

R IZ = Vt lP corc
The cur rent drawn by the primary winding at no-load
is the sum of the active current J Oa and t he reactive current t.;

i, = i.; + i; = V11R 12 + TT1/j X 12 =

V1Y o


The admitt ance

is equivalent to R IZ and jX 1 2 connected in parallel.
Equation (2-25) describing the events occurring in a transformer at no-load, with allowance for the core loss, corre sponds to the equiva lent circuit shown in Fig. 2-7a. In practical cal cul ations, howe ver, it is more convenient to re-draw








jX 12






(a )

Fig . 2-7 E quiva lent circuit of a transfor mer on no-load :

(a) wit h R an d X connect ed in p ar allel; (b) with R and X connecte d in


the equivalent circuit as shown in Fig . 2-7b whi ch includes

t he prim ar y imped ance at no-lo ad




+ jX o


On expressing Zo in terms of Y o in Eq . (2-26)

Zo = R o + jX o = 1/Y o = 1/(l IR 12 + 1/j X 12)

and equati ng the coefficient s of the im aginary an d real
par ts, we fi nd t hat
Ro - .R ~1 2 X 12
I(R 2t e -~1- X21 2 )
X o = X12R ;2/ (R~ 2 + X ~ 2 )

The loss across R 12 a t VI is equal to th e core loss.


Ch. 2 Pro cesses in Transformer at No-Load

becau se when I oa l or> it is inevit able that R 12 ~ X 1 2 .

Fina lly , we may write

n, = X ~ /R12 ' x, = X 1 2 , R o R 12
The quantity X 0 retains the name of the mutual inductive
reactance of the primar y winding. R o is a fictitious res istance the loss across which at l ois equa l to th p 1',(>1'0. loss of
the t ransformer
P ear e = I~R o

As is seen from the equivalent circuit in Fig. 2-7b, the

products Rol o and jXol o are, res pectively, the active and
reactive components of the primary voltage

T\ .

If 12

Fig. 2-8 Ph asor dia gram of

a transform er on no-loa d

F ig . 2-9 Impedance of the equi "

valent circuit and no-lo ad cur ren t 1 0 as functions of TTL

The relation hetween the pr imary voltage Tll and the

no-load current 1 0
Tll = - E 1 = Z oI o
is illustrated b y the phasor diagram in Fig'. 2-8 which, with
P ear e = 0, R 12 = 00, and lo a = 0, is the same as that
shown in Fig. 2-6 .
Because the magnetic circuit of a transformer is nonlinear ,
the no-loa d current 1 0 rises at a faster rate t ha n VI' so R o
and X 0 depend substan tially on 111 (Fi g. 2-9):


X 12


Vl /l o


Part One. Transformers


R o ,.... (V)I o)2

In contrast, as the primary voltage is va ried, R 1 2 remains
practically unchanged, because the core loss is pro portional
to the square of the magnetic induction, Eq . (2-21), or the
primary voltage, Eq . (2-20).

Electromagn etic Processes

in the Transformer on Load


The Magnetic Field in a Transformer

on load. The MMF Equation.
The leakage Inductance
of the Windings

When a transformer is operating on load, i ts secondary is

traversed by a current

The load current gives rise to a change in the primary current. Proportionate changes also occur in the magnetic
flux and the secondary voltage , aud thera is an increase


Ii.I ~f ~
p/ -





!tZtt62 Z
- Pz

Fig. 3-1 Single-phase, two-winding transformer



in the power lost. For a proper estimate of these changes

in a transformer on load , it is essential above all to examine
its magnetic field and to deve lop voltage equations for its
pr imary and secondary windings.

Ch, 3 Processes in Transformer on Load


Figure 3-'1 shows a single-phase , two-winding transform er

whose second ar y is conne cted acros s a load impedance Z.
Assuming that all the relevant electric and magnetic quantities are var ying harmonically , we may write them in complex notation . In doing so, it is important to remember that
the instantaneous value of a harmonic quantity is t o be
construed as the re al par t of t he respective complex amplitud e multiplied by exp (jwt)

= Re [V 2 j exp (jwt)]

[V 2Vexp (jwt)]
e = Re [V2Eexp (jwt) J

v = Re

= Re [cP m exp (jwt)J

lJf = Re [O/m exp (jw t)J


The adopted positive directions of the abo ve quantities are

shown in Fig . 3-'1. Positive directions for II and 1 2 ar e chosen
such that they set up a positive mutual magnetic flux . Posit ive directions for t he voltages and emfs across the win dings
are the same as for the res pective curren ts . Positive diractions on load are chosen the sam e as for oper ation at no
loa d.
When a transform er is operating on lo ad, its magnetic
flu x is established by the prim ar y current II traversing the
primary winding and by the second ar y current 1 2 traversing
the secondary winding . To simplify the matters , this magnet ic flux can be visualized as a superposit ion of two fluxes,
namely the mutual (or m agn etizing) flux and the leakage
flu x .
Th e grea ter proportion of the flux linking the windings
is the mutual flux whi ch h as all of its pa th within the core
and com pletely encloses the wind ings fr om both sides.
The mutual flux cD (Fig. 3-'1) is the same at an y section of
the core ; it s linkage with t he prim ary is WlcPnll and with
the secon dary , w 2 cD m . Und er Amp ere 's circuital law, the
magnetic intensity du e to mu tual induction is t he sum of
the prim ary and seconda ry mmfs


Part One. Transformers

Since t he mutual magnetic induction and the mutual

flux are connected to the field int ensity in a well -defined
manner (see Chap . 2), we may arg ue that the mutua l flux <l>
is established by the sum of t he primary and secondary
mmfs. This sum may be vis ua lized as the mmf due to some
current i o traversing the primary winding
ilWl + i 2w2 = iOw l
Therefore, the current given by
i o = (ilWl

+ i 2w2)/ W l

may be called the mag netizing current, and Eq. (3-1), an

mmj equation .

The non linear effects taking pl ace in the transformer core

as it undergoes cycles of magnetization by the current i o
may be accounted for as in t he case of no-load operation.
The nonsinusoidal current i o may be rep laced by an equivalent sinusoidal magneti zing current the rms va lue of which is


IBa + I ar
and whose active component l oa is related to the core losses.
Then we may write the mmf equation in complex notation as

I lwl


I 2w 2

I owl


In our further discussion, the term "magnetizing current"

will refer to the equivalent sinusoidal magnetizing current 1 0 ,
Now we are in a position to present the primary mmf i l w 1
as a sum of iOw l and (illVl - iolVl) = -i 2lV z which ba lances
the secondary mmf i zlV 2 , and the magnetic flux in operation
on load as a sum of t hree fluxes, namely:
(a) the mutual magnetic flux <l> and the leakage flux
with flux li nk age 1Jf aO' set up by the primary mmf iolV l
(Fig . 3-2a);
(b) the leak age flux established by the mutually ba lancing mmfs, namely (illV l - iolVl) = - i zlV z on the primary
side and i zlV 2 on the secondary side (Fig . 3-2b).
Referring to Figure 3-2, it is seen that the lines of the
leak age flux have their path comp leted through nonmagnetic
(air, oil, etc .) gaps alb l and a zb2 comparable in leng th with
the portions of the lines accommodated within the core
(bla l and bza z) . These lines link either the primary turns
(1Jf o i and 1Jf ao), or the secondary turns (1Jf az).


Ch. 3 Processes in Transformer on Load

The lines of the leakage flux in a transformer may be

divided into two groups-those linking only the primary
turn s and giving rise to the flux linkage 1Jf 00 due t o i o and








iri o


w, t










~ li,""wl



Fig. 3-2 Magnetic flux on load as the sum of (a) mutual flux and (b)
leakage flux

1Ya1 due to (i1 -

i o) , and those linking only the secondary

turns and giving rise to the flux linkage 1Jf 02'
To appraise the relationship between the flux linkages
and the currents in the windings, we shall develop an equat ion by Ampere's circuital law for, say, a closed line of
the leak age flux linking t he primary winding as shown
in Fig . 3-2b:

~ H dl =




Jr n, dI + ~I\ H eor e dl = (ii- i o)


Let us write the magnetic field in the nonmagnetic region,

H 0 ' and the magnetic field in the core , H eore , in terms of
the respe ctive induction and permeability:

H eoro
~[. a. e ore


B oht o

B eore/~t a , eo re
~t r . eore ~t o

~r,eore ~ '1

Therefore , the leakage field ill the core is negligibly small


B eorehta ,eore =


Part One. Transformers

The total current is equal to the magnetic potential difference across the nonmagnetic gap


:0 ) n, dI

) n, dl =

(ii - io) Wi

It follows from the foregoing that 1.J.! o i is proportional

to (i 1 - io). The same holds for 1.J.! co and 1.J.! cr2 and their
respective currents i o and i 2 Therefore, the leakage inductances of the windings
L cr1 = Wcr1/ ( i 1 - io)
t.., = 1.J.! cr2/i2
L cro = 1.J.! cro/ i o
are constant for a gi ven transformer and solely depend 011
the wid th of nonmagnetic gaps and the number of turns in
the windings (see Sec . 8-2) .
With a high degree of accuracy, the total leakage fl ux
linkage with the primary winding may be written
Wcrl = 1.J.! co

+ 1.J.! o o =

Lcroi o

+ L cr1 (i 1-io) ~ L cr1i1


because in operation on load i 1 ~ i o, and we may neglect

wh atever diff erence there may be between L cro and L cri
and deem t hat L cr o ~ L cr1. By analogy with the mutual
inductance [see Eq. (2-16)J, t he leakage inductances may
be expressed in terms of the respective permeances, A cr i
and A cr 2:
L cr1


L cr 2

= w;A U2

or in terms of perm eance coefficients

L cr2

= f10 Wi Acrl
= ~LoW~Acr2


= A u!hlo

L cr1

wher e

Acr2 =


A cr2/ f1 0



V o ltage Equ ations

of the Transfor mer Windings

The emf induced in each of the transformer windings can

conveniently be presented as the sum of the mutual emf E 1
(or E 2 ) and of t he leak age emf E U 1 (or E cr2).


Ch. 3 Processes in Trans form er on Load

The mutual flux shown in Fig . 3-2a does not differ from
that in a transformer OIl no-load (see Fig . 2-1) . Th erefore,
the mutual emf ma y he expressed in terms of the mutual
flux in precisely the same manner as at no-load .
Given a certain E I , the magnetizing current i o must be
the same as at no-load, provided that E I and CD are the same
in either case . Therefore, 1 0 and E I can he conn ected by
an equation of the form




+ jX o

Using the turns ratio, n 21 = w2 /w ll we can wri te the mutual emf on the secondary side as

The primary and secondary leakage emfs, e crl and e cr2 ,
are induced by the leakage flux linkages "If o i and 'P" cr2,
respectively, proportional to the primary and secondary
currents :
e crl = - d'P" crl/dt = -L crl dil/dt
e cr2 = -d'P" cr2/dt = - L cr 2 di 2 /dt
- E2

- n 21E I

n 21Z0IO

Using complex notation and differentiating by analogy

with Eq . (2-19), we get
and, similarly


are called the leakage inductive reactance of the primary and

second ary, respectively.
As is seen from Eq. (3-'10), the leakage emfs ar e in quadrature lagging with the associated currents.
Now that we have defined the primary and secondary emfs
of a loaded transformer and recalling that all the quantities
involved vary harmonically*, we ma y write Kirchhoff's

* The nonsiuusoidal magnetizing curren t is replaced b y an equivalent sinusoi dal cur rent .



Part One. Transformers

voltage equati ons for t he prim ar y and secon dar y windings

ill complex form as


+ u,
= RIll
. .
E 2 + E 0 2 = R 2l 2 + 11



where R, and R 2 are the resistances of the primar y and

secondary windings, respe ctively, including add ition al losses
due to altern at ing current (see Sec. 31-2).
In writing Eqs. (3-12), positive directions were chosen
as shown in Fig. 3-1. The volt age 111 is the supply emf
impressed on the winding from an external source. The
voltage 11 2 = Z l2 is the voltage drop across the l oad on
the secondary side with an impedance of value Z =
= R
jX . Expressing the leakage emfs in (3-12) in
terms of the respective leakage induct ive reactances and
currents (3-10), we may re-write t he voltage equations as

11 2 = E 2 - l2 Z2
where Zl = R,
jX l and Z2 = R 2
jX 2 are the complex
imped anc es of the primary and secondary windings, respectively .


Transferring the Secondary Quantities

to the Primary Side

The performance analysis of a transformer can greatly be

simplifi ed, if we transfer the qu an tities associated with the
secondary to the primary winding. This technique consists in
tha t the real t ransformer having in t he general case different
numb ers of primary and secondary turn s, W I and W 2, is rep la ced by an equivalent transformer in which the secondary ha s
the same number of turns as the primary, w~ = WI (see Fig .
3-3). The qu an tities associated with the equivalent secondar y
ha ving WI turns are said to be transferred (or referred) to
th e primary winding or side. They are expressed in terms
of the original secondary quantities adjusted in value by a
suita bl e factor so that transfer of secondary quantities to the
primary side wi ll leave the magnetic fie ld, and the power
fluxes PI'
P 2, and Q2 unaltered . The procedure is as
follows .



Ch. 0 Processes in Trans former on Load

('1) To leave the ma gnetic flux (I) unaltered , we must

retain the secondary mmf unchanged, that is

I~ Wl =

I zw z

whe nce

j~ = j ZWZ/Wl


Here and elsewhere , the prime on a secon dary quantity

in di cates that it has been tr ansferre d to the prim ar y side .
(2) With (I) ke pt constant, t he emf is proportion al t o
the t urns number. Therefore, the emf acro ss the secondary







Pf -


jE! It;,

z.~ tv;

- Pz

Fig. 3-3 Transformer of Fig . 3-1 with i ts secondary transferred to

= WI
the primary,


winding transferre d t o the primary side will in crease W 1/W2

E~ = E Zw1/w Z
(3) To keep unchanged the values of P z an d Qz drawn by
the load on t he secon dar y side, its R an d X mu st be repl aced by th ose t ra nsferred t o the primar y side:
P 2 = RI~ = R'I~2
Q... = XI 22 = X ' I'22
Using Eq. (3-14), we get
R' = R (W 1/WZ)2
X ' = X (w1h v z)2
Therefo re,
We can see th at the secon dary impedance can be transferre d
to the prim ary side, adjuste d in value by t he turn s ratio

Part One. Transformers


The secondary voltage call likewise he transferred to th e

primary side, adjusted in value by the turns r a tio

V; = Z ' j~ = Z (w]/w z)z zWz/w] = VZw]/w z
The secondary impedance Zz, its resistive component Hz
and its inductive component X z can be transferred to the
primary side in about the sam e manner:
Z; = H;
jX; = Zz (w]/wz) Z
H; = Hz (w]/wz)Z
(w]/w z)Z
As a result, the secondary voltage equation takes the
E~ = Ez (w]/w z) = VZw]/w z Zz (w]/wz)Z jzw z/w j




V'2 + z.i:
2 2

Because the primary and secondary windings have the
same number of turns, the transferred (or referred) secondary
emf is the sam e as the primary emf :


E; = E 2w]/w Z = E]
The mmf equation for a transformer with its secondary
parameters transferred to the pr imary side is ex tended to
include the secondary mmf expressed in t erms of the secondary current referred to the primary wind ing

. lOw]
Dividing the above equation through by
equation of transformer currents


give s th e

I ] + I ; = 10
which has the same physical meaning as the mmf equation
(2-30) . With a suffi cient ly heavy load, when the primary
current markedly exceeds the magnetizing current, I] ~ 10'
th e current equation can approximately be written as

I] = -I; = - I zwz/w]


I] /I z = wz/w]
As is seen, given a heavy load, the referred secon rlary cur rent, I;, does not differ from the primary current , I ].

Ch. 3 Processes in Transformer on Load



The Phasor Diagram

of a Transformer

The voltage and current phasor diagram of a transformer is

a graphical interpretation of the equations describing the
performance of the transformer. These equations includ e
- the winding voltage equations



-E I

-E~ =



-EI =

-the load voltage equation


= Z'

+ Z~ (-j~)




-the mu tual emf equation

- EI




-the current equation


i, -



Using a ph asor diagram constructed to a cer t ain definite

scale, we can determine the voltages, emfs and currents of
a transformer on load . The sequence in which a phasor
diagram is constructed depends on which quantities are
specified to define the operation of the t ransformer and
which quantities are to be determined .
Suppose that we know the secondary current 1 2 and the
load impedance Z = R
jX (for an inductive load , X > 0;
and for a capacitive load, X < 0). We set out to find the
secondary voltage 112 , the primary emf E I , the magnetizing
current 1 0 , the primary current II' and the primary voltage VI' Th e phasor diagram is usually constructed for
the transformer with its secondary quantities referred to
the primary side. Therefore , the first step is to determine
the secondary quantities referred to the primary side (that
is, adjusted in value by the turns ratio or the t urn s ratio
squared). The referred secondary current is

I~ =
;;- OI6D


(W~/W l)

Part One. Transformers

and t he referred imped ances ar e


Z~ =

Z (WI/W2) 2 = R'

Z 2 (WI/W 2)2 = R~

+ jX'

+ jX~

The ph asor diagram is m ade more compact if the complex

quantities referred to the primary side are t aken with a
minus sign , -1~ an d
The first to be pl otted (see

- r;

Fig. 3-4 Pha sor dia gram of a tra nsformer operating int o a resist iveinductive loa d (CP2> 0, X > 0)

Fig. 3:.. 4) should be -1 2 which ma y be drawn in an ar bitrary

direction , say along t he positive axis of the complex t ime
pl ane and on the sca le adopted for currents. Th en , using
t he load voltage equation, we find the referred secondary
. voltage: - V~. This voltage has an activ ~ component,
R' ( - 1~), an d a re act ive component , j X' ( - 1~ ), wh ich are
laid off t o the adopt ed sca le . The active comp-onent is la id
off in the direction of -1~ , whereas the reac ti ve com ponent
leads -1~ by 90 if the load is inductive and X > O. The
actual second ary volta ge is found by Eq . (3-17):


Vi (W2/Wl)

Ch. 3 Proc esses in Transformer on Load

Then we find graphically the mutual emf -E 1

and compute the magnetizing current
10 = E 1/ V Rfi



+ X~

and the ph ase angle

CjJo =


(XolR o)

Now j 0 can be laid off on the phasor diagram . The mutual

flux cD can be found from Eq . (3-7) and laid off on a scale
of its own (the flux is in quadrature lagging with -E1) .
The primary current II is deduced from t he current equation .
The primary voltage VI is found graphically in a similar
way. The construction thus obtained also gives the phase



Fig . 3-5 Ph asor diagra m of a tr ansformer operating into a resi stivecapaciti ve load (rr2 < 0, X < 0)

shift (P2 between t he secondary voltage and current, and

the phase shift (PI between the corresponding primary quantities.
With a resistive-inductive load , both the primary and
the secondary currents lag beh ind the respective voltages
in phase, so CjJl and CjJ2 are taken to be positive: CjJl > 0
and CjJ2> 0 (see Fig. 3-4).
The ph asor diagram for a resistive-capacitive load is
plotted in Fig. 3-5. As is seen, the secondary current leads
the voltage by an angle (P2 (CjJ2 < 0). If the load is predominantly capacitive (see Fig . 3-5), the primary current like5*


Pa rt One. Transformers

wise leads the voltage by an angle (PI < O. If the capacitive

component is less pronounced, the primary current may
even lag behind the voltage .

The Equivalent Circuit

of the Transformer

If we treat a single-phase, two-winding transformer as

a two-port, the equivalent circuit stems from Eqs. (3-22a)
through (3-22d), where the secondary quantities are trans-

ferred to the primary side .

Given VI' the circuit equivalent to a given transformer
must draw from the supply line the same primary current i,
as the transformer itself. In order to identify the configuration of this equivalent circuit, we must express the
primary voltage in terms of the primary current.
To begin with, we shall express i, in terms of E1 and the
circuit parameters

l/Zo+l /(Z~+Z)

Substituting the above expression into the voltage equation


Vi =

j1 Z1 -E 1= i, [Z1 +-


+ 1I ~Z~ +Z/)

] =

j1 Z eQ


It is seen from Eq. (3-23) that the transformer equivalent

circuit drawing a primary current II must have an equivalent
impedance given by
ZeQ = Z1

+- 1/Zo +1I\Z~+Z/)

This impedance is presented by the circuit in Fig . 3-6 where

ZI is shown connected in series with a parallel combination
of Zo and (Z;
A detailed analysis would show that the individual arms
of the equivalent circuit carry the same currents as the


Ch. 3 Proc esses in Transformer on Load


windings of the transformer in whi ch the secondary quantit ies are tra nsferred to the primary sid e . Also, the current s

Fig . 3-6 E quivalent circ uit of a t ransformer

ent ering the nod es of the circ uit and it s loop voltages satisfy
the basic tra nsformer equat ions.

The Per-Unit Notation

E lectrical quantities (such as currents and voltages) and

circuit paramet ers (reactances an d resistances) can be expressed each as a fraction of an arbitrarily ch osen base or reference quantity , thereby gi ving per-un it quantities.
The per-uni t notation sim plifies the equat ions describing
t r ansformer performance. It also simplifies a chec k on the
design data an d result s , because the per-unit qu an titi es of
different t r ansformers differ much less t ha n the same quantit ies expressed in absolut e units .
The base quantities usually chosen for the primary side
of t ransformers are :
-the rated phase primary voltage, VI, R
-the rated phase primary current, I I. H
- the rated impedance presented by t he t r ansformer t o th e
supply line,
I Z1, n I = VI , nlII, n
-the power r ating of t he t r ansformer

8 1, H = VI, RII , R
in the case of a single-phase transform er , and
81, H = 3V I , RII , R
for a three-phase tr ansformer ,


Part One. Transformers

The base quantities usually chosen for the secondary side

-the rated phase secondary voltage, V z, R = VI, R (WZIWl);
-the rated phase secondary current, l z , R = 11, R (wl/w z);
-the impedance presented to the line on the secondary side
(at V z, Rand i ; R)
I z; R I = V z , RlI z, R =

I z; R I (wZlwl)Z


-the base power on the primary side

S2, R = Sl, R
To obtain a per-unit quantity on the primary side, its
absolute value is divided by an appropriate base quantity
taken in the same units

VI: l

Vl/V l , H

1,!:l = lIllI, R

I Z:I:O I = I z, III z; R I
I Z*l I = I z, III z; R I
P:I: l = Pl/S], R = Vi:l1,r.l cos


where an asterisk stands for per unit. Sometimes, this index

may be omitted, if the use of the per-unit notation is referred
to in the text. The power equation in per-unit quantities
is equally applicable to single- and three-phase transformers.
The quantities associated with the secondary winding
of a transformer can be expressed as per-unit quantities in
anyone of two ways. For example, we may divide a given
secondary quantity taken in absolute units by the corresponding secondary quantity taken as the base. Alternatively
the secondary quantity may first be referred to the primary
side by adjusting it in value by the turns ratio or the turns
ratio squared, as the case may be, and the result may then
be divided by the adopted base quantity associated with
the primary side:

V 21V2 , H = V~fl1l, R
1,1:2 = 1 21I 2 , H = I~lIl, R
IZ:,:21 = I Z21/1 Z2,R I = IZ~ I/IZl,R I
P*2 = P 2/S 2, H = P~/S l, R = V:J: 21,r. z cos crz

For obvious reasons, the secondary quantities expressed on

the per-unit basis carr y no referring index,


Ch. 3 Processes in Transformer on Load

Anyone transformer equation may be written in per-unit

notation . . To this end, it must be div ided through by the
corresponding base quant it y . As an exam ple, let us do this
for Eq. (3-13) which gives t he primar y voltage


V:I: I = -E:':I

+ Z:I:lI,1:l


For the current equation, we obtain



+ 1*2 =



As is seen , the per- unit equations are written in about

the same way as those in absolute quantities, except that
they have no indexes to show transferring to the primary
Per-unit quantities are also helpful in expressing the
parameters and quantities involved in equivalent circuits ,
and in constructing phasor diagrams . The per-u nit parameters and losses of a transformer va ry within a ralative ly
narrow range of va lues and depend ma in ly on it s power rating . Let us establis h the relations between some of the perunit quantities . Among other things, we will find that the
mutual inductive reactance varies inversely as the no-load

X:I: O = I Z:I:O I = I z, II I ZI, R I

= (TTl, RII 0) (II, RIV I , R)
= I I, RlI o


The resistance during magnetization can be expressed in

terms of the no-load current and the core losses (the no-load
or open-circuit losses) as

n., =

Roll z; R I = P coreII, R/3I~Vl,

(Pcore/3VI. RII, R) (II, RlI o)2

~ P:~, corel I~:Q


Part One. Transformers

Finally , the winding resistances are equal to the copper losses

R :I: 1 = RIll z; R I = 3RII~, R/3T1 1, RI 1 , R

= PCu, /8 1, R = P:"cu, I
R:1: 2 = R~/I Zl,


I = 3R~I~, R/3VI , nIl , Ii

Pc, 2/ 81, R = P:1: ClI ,

Using the above relations and data sheet values, the range
of values for the basic per-unit quantities of three-phase
power transformers rated from 25 to 500 000 kVA can readily
be defined. Transformers with higher ratings have lower
resistances and higher inductive reactances:
I,~o =

P:':l, CU

0.03 to 0.003

= P*o = 0.005 to 0.000 6

P:I', core
P:1' 2, ClI


P:I:, cu = 0.025 to 0.0025

0.03 to 0.07
I Z:I:O I = X:I,o = 33 to 330
R:I: I = R u = 0.012 5 to 0.001 25
R,~o~= 5.5 to 65
As is seen from the above figures, as the power is changed
by a factor of 20 000, the per-unit quantities change not
more than ten-fold (in fact, X*l and X:I: 2 only change by
a factor of 2). As can readily be checked , the same parameters
expressed in absolute units will change by a factor of many
hundred thousand .

The Effect of Load Variations

on the Transformer

In a transformer, the primary and secondary windings are

coupled by a mutual flux . Therefore, any change in load
impedance (the impedance on the secondary side), with the
primary voltage held constant, leads not only to a change
in the secondary current, but also to a change in the magnetic
flux, the magnetizing current, the primary current, and the
secondary voltage. After the transients associated with
a load change die out, the transformer settles down to a new
steady state in which the electric and magnetic circuits are
at equilibrium . In other words, the currents in the windings
and the magnetic flux in the core take on va lues which again

Ch, 3 Proc ess es in Transformer on Load


satisfy the conditions of equilibrium for its electric circuits

defined by the voltage equations, (3-13) or (3-19), and for
its magnetic circuit define d by t he current equ ation (3-20)
supplemented by the emf equations (3-7) and (3-8).
A change in the secondary current immediately brings
about a change in the peak magnetic flux cD m and the primary
emf E l it induces. The state of equilibri um that exist ed
on the primary side prior to that change and with which
was associa ted a certa in definite pr imary cur rent is ups et,
and a current is induced in the primary in accord with
Eq. (3-13)


n\ -

(-E\)] /Zl

The primary emf and the primary current keep varying un til
t he magnetizing current (with the new value of 1 2 ) and the
corresponding emf, -E l = ZoI o, build up enough for a
st ead y-state current to appear in the primar y winding.
Considering together the equations written earlier, the
primar y current (Fig. 3-7) ma y be written

t. = l\.
R/(Zo + Zl) .

10 NL - I~Zo /(Z o


1 0 NL


+ Zl)

+ Zr)

VI, R/(Zo


+ Zl )

is the ma gnetizing current at no-load .

Because Zl ~ Zo, with a sufficiently large load we have
1 0 NL ~ 1




The magnetic flu x varies directly wi th E, which is in

t urn a function of t he magnitude and phase of the primary


-E l

or, in per-unit

VI. R - ZlI l


-E*l = V*l . R - Z*l!,l:l

At no-load , when I I = 1 0 NL ~ 0, the emf and the flux
are equal to the prim ar y voltage taken as unity

eon =

CD n = cD ~/cD ~ l R =

V'!'l, R

= 1

Part One. Transformers


At rated load (I*1 = 1*1, R = 1), the emf and fl ux change

in significantly in comparison with their no-load' values.
Ev en when t he phase of I I , R is such that ZlI1 is in the same
or oppo sit e direction with VI , R, the emf is

where Z','1 = 0.03 to 0. 07 (see Eqs. 3-33).

Thu s , even in t he wor st loading case, with the load rising
fr om zero t o its full value (see Fi g. 3-7), t he emf an d flux
change by as li t tle as
Z,!:1 X tOO = 3 %
Given other ph ases for 1,
an d 1 2 , the changes in the emf
an d flux are still more insignificant . Referring to Figs. 3-4
and 3-5, t he emf decre ases in
the case of a resistive-ind ucti ve
load an d m ay in crease if
t he lo ad is resistive-ca paci t ive
~====~~~';I2~ and the ph ase sh ift is cl ose
t o - n/2. The effect of lo a d
var iations on the m agnetizin g
current is likewise insignifiFig. 3-7 Flux , primar y emf,
cant. It can be eva luate d bv
magnet izin g current and priE q . (3-20) :
mary current as functions of


secon dary cur rent: soli d line ,

resistive-inductive load , CP2> 0;
dashed line , resistive-capa citive
load , CJl:i < 0; Io x=Io.NL

i o = - E1 /Z o


= (VI - Z l I 1)/Z o

=1 0 ,

NL -

I 1Z 1 /Z o (3-3G)

In a linear approx ima t ion , this cur re nt varies in the

same m an ner as t he pr imar y emf. If we in clu de t he nonlinear beh aviour of t he m agne ti c circuit whi ch causes Zo
to vary as well , this cha nge bec omes more pr onounced . The
effect of nonlin earity m ay be account ed for by using t he
ma gnetizat ion curve, (J) = f (I o) , shown in Fig. 2-9.
Plots of II , E 1 , (D , and 1 0 as fun ctions of 1 2 for inductive
and capacitive load s ar e sh own in Fi g. 3-7,


Ch. 3 Pr ocesses in Transforme r on Load


Energy Conversion
in a Loaded Transformer

The energy fed int o t he pr imary winding of a transformer

from a supply li ne is customarily treated as t he sum of two
One pa rt is delivered to load and is partly lost in the
transformer it self. The average time rate of this unidirectional flow of energy is called the active power drawn by the
primary winding from t he supply line. For a single-phase
transformer, it is given by

VIII cos


VI lla



where I l a = I I cos CPI is the active current

VI a = VI COS CPI is the active voltage
The act ive power is taken as positi ve, PI > 0, if CPI li es
anywhere between _90 0 and +90 0 (electrical) .
The other part of in put energy is spent to establish magnet ic fie lds in t he transformer it self * and also electric an d
magnetic fields in the load. The direction of this energy is
changed twice every cycle , so the res pective power averaged
over a cycle is zero .
The transfer of energy between the supply line and a field
(electric or magnetic) is described in terms of the peak
inst ant aneous power , called the reactive power. The reactive
power drawn by t he pri mary winding of a single-phase transformer from the sup ply li ne is given by

VIII sin


VII l r



where I l r = I I sin CPI is the rms value of reactive current

Vir = VI sin CPI is t he rms va lue of react ive voltage
The reactive power is assumed to be posit ive, QI > 0, if
the reactive current is lagging beh ind the voltage,
<:n, wh ich corresponds to a resistive-inductive load. The
reactive power is taken to be negative, QI < 0, if the reactive
current is lead ing t he voltage,
> (PI> - :n , which cor resp onds to a resistive-capacitive load .
Consider the conversion of active power in a transformer.
Let us write the active component of the primary voltage,
VIa = VI COS (PI, as the sum of projections of Eland the

* The energy associated with the electric field wi thin th e transformer is usually ne~lecte~,

Part One. Transformers


voltage drop R]I] (see the phasor diagram in Fig. 3-8a)

VIa = V] cos cp] = E] cos 11J]
and the active power p] supplied to the primary winding
by a supply line (its direction is shown in Fig. 3-9 by an
arrow) as the sum of two components
p] = (11] cos (p]) I] = (E] cos 1h) I]
(R]I]) I]
The term fiR 1 = P C ll , ] is the copper loss in the primary
winding, that is, the power lost as heat dissipated in the
primary turns (see the arrows in Fig. 3-9).
Referring to Fig . 3-8b, the active component of the primary
current, I] cos 1IJ], is shown as the sum of the active components of the magnetizing current locos CPo and of the
secondary current I~ cos 1P2' Therefore, the term (E 1 cos 1P]) I]
may likewise he written as the sum of two components:
E]I] cos 1p] = E2I~ cos 'ljJ2
E]I O cos CPo

P e~

+ P eor e


The term P em = E]I~ cos 11J2 is called electromagnetic

power. It is transferred inductively from the primary
to the secondary winding. The flow of electromagnetic power
crosses the channel between the two windings (Fig. 3.9).
The term E]I o cos (Po = E]I oa = P eor e represents core
loss in the transformer.
Referring to Fig. 3-8c, the active component of the primary
emf, E] cos 1P2' can be expressed in terms of the active component of the secondary voltage, V~ cos (P2' and resistive
voltage drop, R~I~. Hence, we may write
P em = (E] cos 1P2) I~ = (V~ cos (P2) I~
= P2
P CU, 2

+ (R~I~) I~

Some of the electromagnetic power is expended to make
up for the copper loss in the secondary winding,
P CU,2


The remainder,

P 2 = V~I~ cos CP2

is transferred to the load conductively (see Fig. 3-9).
The active power input to a transformer is
p] = P CU, ]
P rpm =

P CU,2

+ P eor e + P em
+ P~



Ch. 3 Processes in 'transforme r on Load


- E,=-Ez


Fig. 3-8 Phasor diagrams of a transforme r oper ating into a resist ive inductive load (see Fig . 3-4)





P~qf )

( k-. .....


>JlT", tl.ll

It. ~ "VI




( ) <Pm

~ -:..c.







Pco;e(Qo) Q6 f

6 2m







~ fm





Fig. 3-9 Flo ws of active and reac tive power in a loaded transformer


Part One. Transformers


2j Ploss = PCu ,1

+ P Cu, 2 + P eore

The conversion of reactive power in a transformer may

be treated in a similar way, likewise referring to the phasor
diagram in Fig. 3-8.
The primary reactive power can be treated as the sum
of the fo.llowing components:

Q1 = (VI sin (PI) II = (E 1 sin tPI) II

= (E 1 sin tPI) II





E 1 (II sin tp1) = E 1 (I~ sin

= Qam


+ E 1 (1

sin CPo)

+ Qo

(E 1 sin tl)2) I~ = (V~ sin (P2) I~

= Q2

+ (X~I~) I~

+ QU2

The physical meaning' of the reactive components is as

Qem = E 1 (I~ sin tp2) is the reactive power transferred
inductively from the primary to the secondary winding
.Qo = E 1 (fo sin (Po) = ElIoT is the reactive power required to establish the magnetizing (mutual) flux
QUI = X II; is the reactive power expended in setting up
the leakage flux 011 the primary side
QU2 = X ~I~2 is the reactive power expended in setting
up the leakage flux on the secondary side
Q2 = V~I~ sin CP2 is the reactive power drawn by the load
The directions of these power components are shown in
Fig. 3-9.

tho it Transformation of 3-phase Currents and Voltages

Transformation of Three-Phase
Currents and Voltages


Methods of Three-Phase Transformation.

Winding Connections

Three-phase currents and voltages may be transformed either

by a bank of three single-phase two-winding transformers
(Fig. 4-1) or by a three-phase , two-winding transformer
whose windings are put on a common magnetic circuit of
the core or the shell-and-core (five-leg core) construction
(Fig. 4-2).
The magnetic circuit of a core-type three-phase transformer
can be formed from those of three single-phase transformers
combined together. Arranging the single-phase transformers
as shown in Fig. 4-3a and combining the limbs that do not
carry any windings (Fig. 4-3b), it can be noted that with
a symmetrical set of voltages (Fig. 4-1), the flux in the
combined limb, equal to the sum of the phase fluxes, vanishes


+ cJ) + CDc

Therefore, we are free to remove the combined limb altogether (Fig. 4-3b). The magnetic circuit thus derived is
sometimes used in practice and is known as a spatial threephase core. In such a transformer, the instantaneous fluxes
in limbs A and C have their paths completed through limb
B, because
CD B = -cJ) A - CDc

Most frequently, however, the magnetic circuit of a threephase transformer is built as a flat (or planar) core-type
structure (Fig. 4-3c), with the limbs arranged to lie in a
common plane. It differs from a spatial core in that the
phase B core has no yokes and the axes of all the phase legs
and yokes lie in a common plane.
A flat core shows a degree of asymmetry which results
in an asymmetry of the magnetizing currents. However,
this is of minor importance because these currents are small.
The shell-and-core (five-leg core) form of the magnetic
circuit is employed in high-power transformers (Fig . 4-2b)
so as to reduce the yoke height. This is achieved owing












b ~






c :>-r-





F ig. 4-1 Three-phase transform at ion by a bank of si ng le-phase transformers






rPy e---:

,- v"b








(CL )

Fig. 4-2 Three-phase t ra nsformers: (a) core- ty pe; (b) sh ell- and-core
(five-leg ra re) t ype




(0 )

Fig. 4-3 Three single-phase magn etic circuits transform ed in to (a)

and (b) a "spatial" core and (c) a plana r core

ell. 4 Transf ormation of 3-Phase Currents and Voltages


to the formation of extra closed paths (through the si(~e

limbs) for the magnetic fluxes . In a shell-and-core m agnetic
circuit, the phase fluxes cD A' cD B ' and cD C may be visualized
as composed of the individual loop fluxes cD u , <Db' and <Dc
(Fig . 4-2b) , such that cD A = d) b- CD u, cb B = cDc - <Db'
and <i:> c = CI)u _. cDc. As follows from the flux phasor dia gram in Fig . 4-2c, construct ed on the basis of design data for

Fig. 4-4 Zigzag-star connection

the core , the fluxes in the yoke loops form a nearly symmetrical star (cD u is somewhat smaller than cD b = cDc), and the
fluxes in yokes band care 1/ V if t imes the fluxes in the
phase limbs. (It is to be recalled t hat in the core-form transformer of Fig. 4-2a the yoke fluxes do not differ from those
in the phase limbs.)
A three-phase transformer is far more economical , so banks
of single-phase transformers are only used where a single
three-phase transformer of t he same power r ating would
have a prohibitively large weight or size .
The phase windings may be connected in a star (Figs . 4-1
and 4-2a) , a delta (the LV winding in Fig . 4-2b) and , though
seldom, a zigzag (Fig. 4-4) . In a star connection,
V ll n e = V A B = V B C = V C A =

JUn e



V 3V p h

Part Dne. Transformers

In a delta connection ,
V ilhe

= Vp h =




V ab

V bc = V ac

In a zigzag (or interconnected-star) connection, the vol tage and current relations are the same as in a star conne ction, but in order to obtain the same phase voltage the number of turns per phase must be increased 21Vf = 1.16 times.
This leads to a higher cost which is to some degree offset
by an improvement in performance (as regards t he waveform
of phase emfs and fluxes) .
The star connection is designated by a "y" sign or the
let t er Y (sometimes spelled in full as "a wye connection") .
The delta conn ection is designated by the Greek let t er f...
or the Roman let t er D. The zigzag connection is designated
by the let t er Z.
A winding carrying the highest rated (line) voltage is
referred to as a high-voltage (HV) winding. A winding
carrying the lowest rated (line) voltage is referred to as
a low-voltage (LV) winding .
In the Soviet Union , the manner in which the windings
of a two-winding transformer are connected is designated
by a fraction , with t he form of connection of t he HV winding
placed in the numerator and t ha t of the LV winding in the
denomina t or . For exam ple, t he form of connection for the
transform er in Fig. 4-2a will be designat ed Y /Y , and for
that in Fig. 4-2, Y IJ],. Outside t he Sovie t Union , the same
forms of connections may al ternatively be desig na t ed as Y-Y
(or wye~wye) and Y-b. (or wye-delta).
Under a relevant USSR standard, the start and finish of
t he HV winding in a single-phase tr ansformer are marked
as A and X, and those of the LV winding, as a and x . The
st arts and finishes of the HV winding in a three-phase transformer will be designated as A, B , C and X, Y , Z , and
those of the LV winding, as a, b, c and x , y, z. The neutral
wire is designated as N , and the centre (or zero) point of
a star connection is marked 0 on the HV side and 0 on the
LV side.

th. 4 Transformation of 3-phase Currents and Voltages


A Three-Phase Transformer
on a Balanced Load

The performance of a three-phase transformer on a balanced

load may be described in terms of the t heory developed for
single-phase transformers. In fact, all the relations derived
for a single-phase transformer fully apply to any phase
formed by a primary and a secondary on a common limb .
Some adjustment needs only to be made for the magnetization of the core in a three-phase transformer (see Sec. 4-4)
and the calculation of the magnetizing current (see Sec. 8-1).
However, the magnetizing currents are negligible in comparison with the load currents, and the unbalance in these
currents related to the dissymmetry of a flat (planar)
three-phase magnetic circuit is of minor importance. Therefore, the actual calculations are based on an equivalent
balanced set of averaged magnetizing currents to which corre spond the averaged mutual impedances (ZOA = Zo B = Zoc)
accounting for the magnetic coup ling between the various
phase windings.
Because the leakage fluxes are concentrated in the space
t aken up by the windings themselves (see Sec. 8-2), the
leakage flux es of the individual phases may be considered
independently of one another, whereas the leakage impedances of the phase windings equal in size may be deemed
identical (X 1A = X 1 B = X 1 C , X 2 A = X 2 B = X 2 C ) . This
also goes for the phase resistances (R 1 and R 2 ) .
Therefore, with balanced primary line voltages and
balanced load impedances, the phase currents and voltages
are likewise balanced. In the circumstances, the line and
phase quantities are connected by simple relations :

= [line

V ph = vline/ -V 3

in the case of a st ar connection , and

V ph = V ll ne

= [line/ Y 3

in the case of a delta connection, and we may describe the '

performance of any of the phases, using the equations, equivalent circuit, and phasor diagram developed for a singleG*




Part One. Transformers

phase transformer (see Chap. 3), extended to include the

phase voltages, currents and impedances, and also the transformation ratio in terms of phase voltages or turns:


V 2 , R(ph/ V1,


W 2/W 1

Phase Displacement Reference ~umbers

For proper use of transformers in power systems, it is important to know the phase displacement between the emfs on
the HV and LV sides, as measured across like terminals. For
example, on the HV side the emf must be measured across
terminals A and B, and on the LV side, across terminals
a and b.
In single-phase transformers, the phase displacement
between the emfs on the HV and LV sides may be 0 0 or 1800
The line emfs on the HV and LV sides in three-phase transformers can only be displaced in phase through an angle
which is a multiple of 30. Transformers having the same
phase displacement between their HV and LV emfs fall
in the same phase displacement (or reference phasor) group,
each group being assigned a distinct reference number.
Since an angle of 30 is exactly the angle between adjacent
hour markings on a clock dial, a convention adopted internationally is to indicate phase displacement as a clock figure
representing the hour read by a clock when the minute hand
takes the place of the line emf ph asor on the primary side
and is set at 12 o'clock, and the hour hand represents the
line emf phasor on the secondary side. The "time" thus read
is the reference number assigned. An example of this convention for phase displacement group 11 is shown in Fig. 4-5.
The positive directions adopted are from A to B and from
a to b.
In the designation of a transformer, the reference number
follows the symbol for the winding connection (for example,
Y/Y-O or Y/ ~-11).
If the phase windings on the HV and LV sides are wound
in the same direction, the LV leads may be marked in any
one of two ways, shown in Fig. 4-6. Because the windings
link the same flux, the emfs labelled by the same letters
will be in phase in case (a), and in anti-phase in case (b).
(As the flux decreases, the HV and LV emfs will be directed

Ch, 4 Transformation of 3-Phase Cur rents and Volt ages


fr om X t o A and from x t o a in case (a), and the LV emf will

be directed from a to z in case (b). )
As already n oted , sin gle-phase tran sform ers can onl y have
zero or 180 0 ph ase displacem ent. Consequ ently , they m ay
bear onl y reference number
(12) or 6, respectivel y . For
brevit y, they are designated
by t he symbols 111-0 (see
Fig. 4-6 a) an d II I-6 (Fig. 4-6 b) .
A cha nge from group 0 (12) to
group 6 call s for n o connection
cha nges in the transforme r
it self; it will suffic e to rem ark lead a as x , and lead x
as a. In the USSR , singleFig. 4-5 Clock-hour conve nt ion
ase tra nsforme rs are ma nto design ate ph ase displ acemen t
ufa ctured with a 111-0 windgrou ps
ing connect ion .
Extending the foregoing to the HV an d LV ph ase windings
of a three-ph ase t ransfor mer an d referring to the pha sor
diagr am, it can be seen t hat a Y/Y three-phase tra nsforme r
with the leads m ark ed as shown in Fig. 1:-7a falls in ph ase

q; t



s: (a)




x >t

(f(6 ) []J

Fig . 4-6 Lead markings and phase displacement gro up numbers

single-phase transfo rmers


disp lace ment group 0, so it s designation is Y/Y -O. (The

phase emf ax is in the direction of the phase emf A X ; t he
phase emf by is in the direction of the ph ase emf BY, cz
-+ CZ, an d the line emf ab is in the direction of the line
emf A B .) If we re-label the leads , going all the way r ound
the circle , we can converL a group 0 trans former to a group 4
or grou p 8 t ran sform er withou t actually sh ift ing any connections in si de t he t r ansfo rme r. Wi th the leads m arked as


Part One. Transformers

shown in the parentheses, (a), (b) an d (c), the line emf (a) (b)
is in the direction of the li ne emf B C (because these emfs
are measured across th e windings put on the sam e limbs),
and the t ran sfor mer is conver ted to one in phase displaceme nt group (4). W ith the lead s labelled as shown in ' the




(e)[b] (a)[e] (b)[a]


(~ (c)

a [a~[e]



(z)[y] (x)[z] (y)[x]

~ --


(e)[b] (a)[c] (b)[a]

(z)[y] (x)[z] (U)[x] [8] [bJ




(e)[b] (a)[c] (b) [a]






. (b)


(z)[y] (x)[z] (y)[x]
.b Z

[a] (a)a
--. (3)
/ A (b)

[7J l [bJ


(e)[b] (a)Ee] (b)[q]

Fig. 4-7 Lead markings and ph ase displacement group numbers for
th ree-phase t ransform ers

bracket s, the emf [ aJ [bI is in the dir ect ion of emf CA , aud
the transformer is converted t o one fa ll ing in ph ase displace ment gr ou p [SI. If we wish to ob t ain a y / y-(j transforme r
(Fi g. 4-7b), we must shift the neu tra l jumper that reverses
the ph ase of all the emfs (the emf ab is in anti-phase with
t he e mf AB) . If we re-la bel the lea ds all the way roun d t he
circle , a group (i translonner will be conve rted to a group
('10) or a gr oup [21 t ra nsformer . (Th e resp ecti ve markings
are given in Fig. 4-7b in paren theses an d brackets, respect-

Ch. 4 Transformation of 3-Phase Currents and Voltages


iv-ely.) This exhausts all the likely even reference numbers

that can be derived for a Y/Y connection.
Odd phase displacement clock numbers are obtained for
a Y /11 conn ection. With the leads marked without parentheses or brackets (a, b, c, x , y, and z in Fig. 4-7c), the line
emf ab which is at the same time the phase emf yb is in the
dir ect ion of the emf YB , and the transformer falls in phase
displacement group H.
If we re-Iahel the leads all the way round the circle as
shown in the parentheses and brackets in Fig. 4-7c, we shall
obtain group (3) and group [7I. (Each time we re-label the
leads, a particular emf is turned through an angle 120
= 4 X 30 and the reference number is in cremented by 4.)
If we in t erchange t he st ar ts and finishes of the phase
win dings, a group 11 t ransforme r will become a group
5 tra nsformer (the respe ctive markings ar e given without
parentheses in Fig. 4-7d). Finally , if we re-label the leads
all the way round the circle as shown in Fig. 4-7d, we shall
obtain group (9) and group [L].
Of all the likely phase displacement groups , three-phase
two-win ding trans form ers of Sovi et . manufacture are only
ava ilable in group. 0 and gro up 11, wit h t he (neutral) lead
of the star available for connection where necessar y (Y/Y n-O,
Y/11-11, Y nll1-11). Additionally , some tra nsformers ma y
have their HV windings connected in MY n-11. As is seen
from Fig. 4-8, the delta connection in this case is obtained
differently than in a Y /11-11 transformer . (A is connect ed
to Z , whereas in a delta on the LV side a was connected
t o y.) If t he delta on the HV side were connected in the same
man ner as the delta on the LV si de in a Y /11-11 connection
in Fig. 4-8c, the M Y conn ection would fall in ph ase displacement group 1, rather th an H.
It is of interest to see how , in the general case, the phase
displacement numb er will ch ange if we make the LV winding
an HV one and t he HV winding an LV one , whil e retaining
their connection and markings.
Obviously , the pha se displacement between the HV and
LV line emfs AB and ab will be the same as before, being
30 X N (Fig. 4-9) . Ho wever , the emf ab in t he diagram
shown by dash ed lines will now lead the emf A IJ b y the
same angle 30 X N by whi ch it lags behind in the dia gram
shown by solid lines. Therefore , if we coun t the phase displacement from emf AB to emf ab always clockwise, the

Part One. Tra nsfor mers


ang le 30 X N ' in the second case will complement the original 30 X N angle to 360
30 X N'

+ 30


Th us, after the above manipulation, the phase displacement reference number N ' can
be found as



N ' = 12 - N

where N is the original clock

For N = 11, N '
figure .
= 12 - 11 = 1.
/1 /Y n-11 connect ion
x y
(N ' = 11) can be derived from
t he Y nl /1-1 connection (N
Fig . 4-8 i1/Y-1J Lra nsfor mer
= 12 - N' = 1) whi ch is in
turn derived from the Ynl/1-11
connection by modifying the
delta connection (see below).
Which ph ase displ acement
(fZ-N) JOo
a transforme r will fall
in depen ds not only on the
sequence of marking the starts
a A
and finishes of its LV winding,
but also on how the phase
windings are connected in a
Fig . 4-9 Using the clock figur e
delta . Under a relevant USSR
and lead markings of th e
LV (BV) winding to deri ve th e
standard, the delta on the LV
clock figure and lead markings side must be formed by confor th e HV (LV) winding
necting lead a to lead y, lead b
to lead z, and lead c to lead x,
as shown in Fig. 4-7 or 4-10 by solid li nes. If, instead, we
form a delta by connecting term inal a to te rminal z, terminal
b t o term inal x, and terminal c to terminal y (as shown in
Fig. 4-10 by dashed lines) , t he LV emf, say ab, will be turned
t hrough 180 - 120 = 2 X 30 clockwise , and the clock
figure will be in cremented by 2. (With the leads marked as
shown in Fig. 4-10, group 3 will change to group 3 + 2 = 5.)
Wi th t he conn ect ion shown by soli d lines, t he li ne emf ab,
which is at the same t ime the phase emf yb , is in t he direction
of t he emf ZC. With the connection shown by dashed li nes,





Ch, 4 Transform ation of 3-Phase Currents and Voltages

t he line emf ab which is now the phase emf ax is in the

direct ion of the emf BY . That is , it is turned from its original direct ion through 2 X 30.
This rule extends to any other odd phase displacement
groups . So , wh en the delta is formed in any other way t ha n
recommended, reference number N will becom e reference
number N = N
2. More specifica lly , inst ead of group 11

a a.





,I ,I




~~ .
b' L~J


Fig . 4-10 Phase disp la cement group resulting from th e manner of

delta connection

there will be group 1; in stead of group 3 th ere will be group 5;

instead of group 7 there will be group 9; instead of group 1
there will be group 3; in stead of group 5 t here will be gr oup 7;
and in stead of group 9 there will be group 11.
A relevant USSR standard recommends that the zigzag
connect ion should solely be used on the LV sid e, and prescribes only one group , namel y Y/Z n-11 , t hat is , one wi th
the neutral li ne at the zigzag available for connect ion .

Yhe Behaviour of a Three-Phase

Transformer During
Magnetic Field Formation

In discussing single-phase tran sformer in Sec. 2-5, we have

seen that when the magnetic flux is sinusoida l , cD
= cD m sin wt, the magnetizing current i o is nonsinusoidal.
In addit ion to the fund amental component, I ol m sin wt ,
varying with an angular frequency oi , the magnetizing


with an

Part One. Transformers

i o contains odd harmonics", 1011, In sin kest, varying

angular frequency kes, where k stands for the

3, 5, 7, 11, '13 , and so on,
i o = lot 'ln sin

wi + ~

101l .1n

sin kwt

Distortion in the waveform of i o increases (the odd harmonics

grow in amplitude) as the magnetization characteristic
CD = f (to) becomes progressively more nonlinear.
In three-phase transformers, the nonlinearity of the magnetization charncteristic lead s 10 far more complex effects,

Fi g. 4-11 H armonic compone nts of a symmetrical set of three-phase

cur rents

and the manner in which they manifest themselves depends

on the type of winding connection and the core design.
Three-phase transformation may be accompanied by distortion in the sinusoidal waveshape not onl y of the magnetizing
currents, but also of the magnetic fluxes and phase voltages.
Before going any further, it appears worth while reca lling
some features of the harmonic components in symmetrical
three-phase systems of emfs, voltages, currents and fluxes .
" Here and else wh ere, only th e re active components of th e no-lo ad
current are considered.


Ch. 4 Transformation of 3-Phase Currents and Voltages

A sy mmetric al t hree-phase syst em h as t hree sets of nonsinuso idal phase quantities (currents, voltages, an d fluxes)
t h at, at an y ins tant, are equal in magnitude , waveform and
fundamental frequency, but are separ at ed in time-phase by
one-third of a period T 1 = 2'Jt/ oi . A symmetrical three-phase
system of nonsinusoidal currents i A , i B, and i c is shown
in Fig . 4-11 .
The fundamental terms of the phase quantities (say, currents) are likewise separated in t ime-phase by a third of
a cycle and form a symmetrical syst em t hat has a p ositive
phase sequence, PPS (Fig. 4-11)

= 1/ 2" 1 A 1 sin wt

iA I

'V21 sin (wt - 2'Jt/3)

=V2 1A I sin (wt + 2'Jt/3)

iB I =

A 1

The sum of phase quantities (say, currents) in t he case of the

positive ph ase sequence is always zero


This can readily be proved if we write t he ph ase qu anti ties

(say , cur rents ) as a symmetrical set of complex qu an tl ties

1 AI'

1 B1

1 A 1 ex p (-j2'Jt/3),

~/C 1 = j A l exp (j2'Jt/3)

Similar symmetrical systems with a positive or negative
phase sequence are formed by all harmonics whose order k
is not a multiple of three (that is , other than triplett harmonics*)
k = 6c + 1
where c

0, 1, 2, 3,

=V 2" 1,lh sin kiot

i B h = V 2I.Ah sin k (wt- 2'Jt/3) =V2I 'l h sin (kw t +
iA h

i C h =V 2 I A h sin k (wt

+ 2'Jt/3) = V 2l


sin (kwt


+ 2'Jt/3)

* Triplen harmonics refer to all harmon ics whic h are mu lt iples

of three.


Part One. Transformers

When k = Gc
1, in whi ch case the upper signs in the
arguments of the sines apply, a positive ph ase sequence
of quantities (say, curre nts) is formed. When k = 6c - 1,
in which case the lower signs in the arguments of the sines
apply, a negative phase sequence (NPS) of quantities (say,
currents) is formed. The sum of the kth harmonics of the
phase quantities is likewise equal to zero

i Ch =
The harmonics of phase quantities , whos e ord er is an
integral multiple of three (triplen harmoni es),




where c = 0, 1, 2, 3 , .. ., form a zero phase sequence system

.(ZPS) . The t riplon-harm onic terms are all in phase
iBh = V2I Ah sin k(wt+2n/3)
(CI ,)

= V ZI Ah sin (kwt) =
i Ah

iB h



i Ch

For the third-harmonic t erms , t his rule is illustrated in

Fig . 4-11.
Now we sha ll see wh at restraints are imposed on nOIJsinusoidal curr ents hy the v arious winding connect ion
types. All harmonic terms other t ha n trip1en , that is , the
Ist, 5th , 7th, 1'11h, 13th, etc. harmonics, form a positive
or negative phase sequence (PPS or NPS) system and exi st
in the phase windings connected in any manner. In the
neutral wire, these harmonics do not exist , he cause their
sum is always zero . The line currents of these harmonic
terms , with th e windings connecte d in a delta , are V 3 times
the phase cur rents, for example

I A B 1 = I BC l = I C A l = V 31.'1.1


where l ABI = I Al - I B1'

All triplen harmonics , that is , those harmonics whose
order is an integer multiple of three , that is 3, 9, 15, etc .,
cannot exist in a wye -connection without a neutral wire
(Fig . 4-12). In a three-phase star- connected system with
the neutral wire , Y n- availahle for connection, the neutral

Ch. 4 Transformation of 3-Phase Currents and Voltages


wire carries a current equal to three times the phase current .

For example
. (4-7)

In a three-phase delta-connected system, the third-harmon ic phase currents circulate within the closed path formed
by the delta, and are not present in the line wires .


3i 3



I iJ ~O I

Fig . 4- 12 Third-harmonic currents in various forms of winding connections

A similar situation exists for nonsinusoidal fluxes in the

various core designs . *
In a three-phase bank of single-phase transformers, such
as shown in Fig. 4-13a, the third-harmonic phase fluxes
cI> AS = cI> BS = cI>cs = cI>s have their path completed within
the core in the same manner as the fundamental terms . The
depe ndence of cI> ~ (l) l
cI>s on i o is re presentable by the
magnetization curve cI> = f (io) (see Fig. 4-13a).
In a shell -and-core (five-leg core-type) transformer, the
outer limbs play the same part as the neutral wire in a starconnected winding. They form . a sp lit "neutral" core which
provides a closed path for the third-harmonic fluxes. In
each outer limb, the third-harmonic flux is 3cI>s/2.
The outer limbs als o provide a closed path for the fundamental fluxes cD AI' (l) B l' and cI> 01 ' Therefore, the depencD 2 on i o may,
dence of the nonsinusoidal fl ux cD ~ (l)l
to a first approximation, be deemed similar to the magnetization characteristic ' of a she ll -and-core transformer with
a sinusoidal fl ux (see Fig. 4-13a).

* In our further discussion, we shall only be concerned with the

fundamental and third-harmonic terms.

Part One. 'I'ra nsformerc

In a throe-phase" core-type '. ,transformer-. .which has no

"neutr al " core in th e form of outer limbs , th e th ird-h armonic
phase fluxes (Fig . 4-13b) have the ir path comple te d thr ough
the tank walls and run in to the appreciable opp osition
pr esen ted by nonmagnetic gaps. Because of t h is, the reluctance seen by the third-h armonic flux es is t ens of t im es
(r:) ep



=o=oU m

,.- --....---

' 'PJ

cf>3 -:


/ / --- -



"""'~- -- -~--







Fig. !i-13 Third -h armoni c flux es in various core des igns

higher than that seen by the fun dament al fluxes traver sin g
a closed path within the core. In determining the fund amental an d third-harmoni c te rms, we have to invoke different
magnetization characteristics . For the third-harmonic flux,
this is the lin eal' magnetizing characteristic, cDs = is (is).
For the fund amental flux, th is is the nonlinear magnetization
cha racterist ic, cD 1 = II (i o), derived for the sinusoidal flu x
upon replacing i o with (i o - is) which gives rise t o the mmf
associat ed with the fundamental flux (Fig. 4.13b)* .

* This is t r ue, if we consi der the fun da me nta l and t h ird-ha rmo nic
terms on ly .

- ,


. '

Gil, 4 Transfo rm ation of 3-Pllase Curr ents and Voltages

Now we shall examine the waveformsof magnetizing -currents, flux es and voltages asso ci ated with the va ri ouswin ding connections an d core designs, assuming t hat at no -l oad
the transformer is energized from the HV side .
1. A three-pha se hank of single-phase tr an sf ormer s .
M Y connection. With the supply voltage impressed on the
delta-connected HV side , the phase voltage is the same as
t he sinusoidal line voltage . Therefore, all t he single-phase
transformers in the bank are connected t o carry a sinusoidal
voltage , an d t hey are magnetized in the same manner as
an individual single-phase t r ansforme r is magnetiz ed with
a sinusoidal volt age (see Sec. 2-5) . In other words, the flux
varies sinusoidally and the magnetizing phase current, nonsinusoidally. The m agnetiz in g current has t he waveshape
shown in Fig . 2-4.
The line conductors carry harmonic currents whose ord er
is no t a multiple of three (esp ecially , t he fu n damental
term i OI ,lI n e) ' Their rms values are V:3 tim es the rms values
of t he phase quantities

o1,llne =

VS l o1

[see Eq . (4-6)1. The t riplen h armonics (especi ally i o3 ) t ra verse a closed path within the delta, and ar e no t present
in the line con duc to rs (see Fig. 4-12). Bec aus e t he ph ase
fluxes con t ain sol ely the fun damental t erms (<3) AI ' <J) E I ,
and cD GI), the foregoing fully applies to dolta-wye-connected
t hree-phase transformers of both t he sh ell-and-cor e an d
th e core t ype .
2. A three-pha se hank of single-phase transformers.
y -Y and Y-Ll connections. If the HV side is ene rgized with
sinusoidal line voltages, the phase volt ages may contain
both the fundamental t erms and t r iplen harmonics, whereas
any other odd harmonics (say , t he 5th , 7th , etc.) cannot be
present in the phase voltages, because t h ey would then be
present in the line voltages as well.
The magnetizing currents in a wye connection with t he
neutral isol at ed (see Fig. 4-12) m ay con tain all harmonics
except the t hir d . If we neglect all harmonics except the
fundamental term , we may , with a sufficient degree of
accur acy , deem that the magnetizing current is a sinusoidal
one , i o ~ i OI (Fig. 4-14) . The m agne tic fluxes in a bank of
single-phase t r ansformers (see Fig. 4-13) contain only t he


Part One. Transformers

fundamental flux and the third-harmonic flux, <:t> ~ <:t>1

-+ <:t>3 (if we reca ll that they will not induce the 5th and
7th harmonic phase voltages) .
The waveform of the magnetic flux is determined graphically , using the magnetization curve <1) = t (i o) for a sinusoidal magnetizing current, i o ~ i 01 ' (Actually , this current


Fig . 4-t4 Harmonic compon ent s of curren t, flux and ph ase emf in a
hank of Y/Y single-phas e t ransformers

contains the 5th and 7th harmonics, because these terms

are not present in the flux.)
The phase flux in Fig . 4-'1 4 is determined to within the
t hird harmonic. As is seen , t he flux waveform is heavily
fl at tened. This leads to a distortion in the sinusoidal waveform of the phase emfs and voltages . With a fla t tened flux
waveform , the phase emf has a well-defined peak (see
Fig. 4-14) which may exceed the fundamenta l peak by as
much as 60 % to 90 %. Accordingly, the t ransformer insulation must be designed for this peak, and this leads to
a more expensive transformer. This is the reason why the
Y/Y connection is not used for banks of t ransformers or where
the magnetic circuit is of t he she ll -and-core form.


Ch. 4 Transformation of 3-Phase Currents and Voltages

Fro m this point of view, it is pr eferable t o use th e M Y

or Y I ~ connection. If such a transformer is energized on the
star-connected HV side, the departure of the fluxes and
phase emfs from the sinusoidal waveshape will be negligible .
With t his form of connection and with an y core design,
the third-harmonic fluxes are reduced by the third-harmonic
currents for which the path is closed within the delta-connected LV winding . The third-harmonic fluxes (Fig . 4-15)




-- -- --~.rA



P"L1 .'







<P3 + eP3A




Fig. 4-15 Damping of third-harmonic currents by the currents circul at ing around a closed delta

induce in the phases of the delta-connected winding the

third-harm onic emfs e3/'. which give rise to third-harmonic
currents i 3~ . Because the delta presents a low (practically
inductive) impedance , the currents lag behin d the emfs
by an angle close to nl2 and set up third-harmonic flu xes
<PM which balance out the fluxes cD 3 almost completely.
3. A th ree -phase, Y/Y, Y/Ym Y /A , or Yn/A-conn ected
core -type transformer. With a Y/Y connection , t he HV
winding energized by a sinusoidal voltage may carry sinusoidal ha rmo nic currents whose order is not a multiple of
three (see Fig. 4-12) . Neglecting all the current harmonics,
except the fundamental term, the current t raversing the
HV winding may be treated as sinusoidal, i 0 ~ i 0 1
In a core-t ype transformer, the harmonic fluxes whose
order is a mu ltiple of t h ree ha ve their pa th completed
through large air gap s (see F ig. 4-13b ). Therefore , given
the same mmfs, they are substantially smaller than in
a bank of sin gle-phase transformers or in a shall-and-core
transformer (see Fi g. 4-1 3a) . Because <P 3 is small in comparison with o, lall the other harm oni c fluxe s ar e negl ect ed) ,
it is legitimate to take it that (]), and <l)3 traversing diffe rent
closed paths do not affect each oth er and are set up independ ently by i o and i o3' resp ecti vely (see Fig . 4- '13 b).
7- 016 n


Part One. Transformers

Assigning some value t o cD I and using the <D 1 = 11 (i o)

curve, we can determine the waveform of i o (Plot ting point 4
for the i o curve is shown in Fig. 4-16. ) As is seen, i o must
contain both i OI and i~3:
i o = i OI
However, a star-connected winding can only carry i OI which
may be construed as the sum of i o and i~3 equal t o -i~3'
that is

The current i o sets up <D 1 , an d io~ = - i~3' which balances

out the third h armonic of i o, establishes cD 3 = 13 (i~3) '

5 '



Fig. 4.-16 H armonic components of curre nt and flu x in a Y/Y threephase core-typ e transform er

As is ~seen from the plot of <D 3 , t h is flux has a sm all peak

<D 3 is distort ed in sh ape
(point 9), so the flux <D = cD I
onl y.Tslightly. Unfortunately, the third-harmonic flu xes in
a Y-Y. t ra nsformer complete their path through the structuraljparts and t ank walls-a fact which might sub st an ti ally
r ais e the no-load losses.
One way to reduce these extra no-lo ad losses and t o improve the wa veform of phase vol t ages, the windings in core-

th. 5 Measurement of Transformer Quantities

type transformers are pr eferably connect ed Y /Y n , Y /I1,
or Ynl 11. Then t he third-harm onic flux es are reduced still
more by t he third-harmonic currents traversing a closed
path around the delta or in the neutral wir e, Y n- The wi nding
losses due to these currents are smaller than those due to
the third-harmonic fluxes in the structural parts and tank
In a delta-connected winding with its neutral wire available for connection, the third-harmonic currents flow in
the line wires and have their path completed through the
transmission-li ne capacitances and the neutral wire. They
interfere with the operation of nearby communication lines
and produce ext ra losses in the cable sheath because, as
follows from Ampere's circuital law, their ex ternal magnetic
field is non-zero.
For this reason, the YIY n-O connection is only used in
small transformers supplying local loads. In all other cases,
a relevant standard recommends to use the Y111-11 . or
Ynl11-11 connection. The YIY connection is not considered
in this standard at all.

Measurement of Transformer


The Open-Circuit (No-Load) Test

The transformer quantities, including losses, can conveniently be measured by an open-circuit (no-load) test and a shortcircuit test.
The performance of a transformer at no-load has already
been examined in Chap. 2. The equations for a transformer
at no-load, wi th all owance for t he primary impedance ZI,
can be derived from the gener al equations (3-8), (3-13),
(3-19), and (3-20), if we set t he load impedance, Z, t ending
to infinity and the secondary current equal to zero

VI = -E 1

+ ZII = II (ZI + Zo)

-V~ = -E~ =






Part One. Transformers

On open circuit, thei load impedance in the equivalent

circuit of Fig. 3-6 must be set to infinity, Z' = 00. Recalling
[see Eq. (3-33)] that ZI ~ Zo, it is legitimate to take it that

= 0 and VI = -E1 = z.i,

The open-circuit test does not call for expensive equipment which would be necessary if the transformer quantities
were measured by tests under load. As its name implies,
the open-circuit test is carried out with the secondary opencircuited, and with the test gear arranged as shown in the


( b) '

(IL )

Fig. 5-1 Measuring the parameters of a single-phase transformer by

(a) an open-circuit test and (b) short-circuit test

test set-up in Fig. 5-1a. The power rating of the variablevoltage source which energizes the primary may be as low
as a few per cent of that of the transformer under test.
During a test, VI is gradually raised from zero to 10 %
above its rated value. Holding the frequency at its rated
value, too, the experimenter measures II = 1 0 (doing this
for each phase of a three-phase transformer) and the power
Po drawn by the transformer under test. Using the data
thus obtained, he plots the no-load phase current 1 0 , the
power Po, and the power factor cos rp = P o/V1I o as functions of the phase voltage VI' In the case of a three-phase
transformer, the plots are constructed for the average phase






and the average phase voltage



+ VB + V c )/3

* We assumed that ZI = 0 already in Chap. 2, because ZI is no

more than one-thousandth of ZOo

Ch. 5 Measurement of Transformer Quantities


Using the values of 1 0 and VI thus found, the experimenter

finds the power factor at no-load, cos ero.
The following transformer quantities are found by an
open-circuit test at the rated voltage. .
1. The transformation ratio defined as the ratio of the
secondary to the primary voltage at no-load
n 2 1 = w 2/wI = E 2,RIE I,R ~ V 2,RIVI,R

2. The no-load current, found either on a per-unit basis

(as a fraction)
or as a percentage


(Io,oeIII,R) X 100%

The no-load current must lie within the limits given in

Eq. (3-33).
3. The mutual imperlance, defined for ZI ~ 0 from
Eq. (5-1),

its resistive component being


P o,oe/31 5,oe =

I z, I cos ero

and its reactive component being

x, 1/ Zfi -Rfi

I z, I sin era ~ I z, I

4. The no-load loss. At VI = VI,R it does not practically differ from the no-load core loss, P eore , De, because
the primary copper loss under these conditions, P eu, I, De
= 315,oe R I , is a small fraction of the core loss, 1 0 , De being
very small.
As has been shown in Chap. 3, the magnetic flux at rated
load remains about the same as it is at no-load (provided VI
is held unchanged). Therefore, given the rated applied
voltage, the core loss at rated load, P eore , is approximately
equal to the core loss at no-load, P cOl,e , DC' and the total noload loss, Po, De
P qore

= P eore, ge = Po




Part One. Transformers

The Short-Circuit Test

In this test, the secon dary is short-circuited, so that the

load impedance is Zo = 0, an d the secondary voltage, V 2 ,
is likewise zero . (In a three-phase transformer , it is presumed
that the secondary leads are all commoned so as to give

0 - - - --+--------' ta )

Fig. 1)-2 Equivalent cir cuit of a transformer on a short circuit :

comple te; (b) simplified


a balanced short-circuit.) The t r ansforme r equations for

a short-circuit test can be derived from the general equ at ions
(3-8), (3-13), (3-19), and (3-20)


VI =




= - i ; = -ZJ; = ZJo
= jo - i;


Using the above equations or t he equivalent circuit in

Fig. 5-2a drawn for the short-circuit t est, we can find the
primary and secondary currents, t he m agnetizing current,
an d mutual emf on short-circuit

L, = Z +Z~Zo/(Z~+Zo)

-t: =
i, =

~e! =

= Vi/Z sc ~ Vii (Zi + Z~)

jiZOI (Z; + Zo) ~ ji


+ Zo) ~ (Tr1/Z o) (Z~/Zsc)

Io , o cZ~/Zs c

zqio ~ zqio. ocZ~/Zsc =



Gh. 5 Measurem ent of Transform er Quantities



= Zl + Z~Zo/(Z~
= R se + jX se


+ Zo) ~ Zl + Z~

+ R;
x ; ~ x, + X;
R se ~ R 1

is the short-circuit impedance, resistance and reactan ce

of the transformer (that is, the impedan ce of a transfo rm er
wi th its secondary short-circuited , as seen by the supply
line) .
The approximate expressions for 11 , I :, 1 0 and E 1
have been derived for 1 0
~ 1 1 , I z, I ~ I ZO I,

I Z ~ I -e; I Zo I [see Eq . (3-33)1,
and are fairly accurat e. Th e
corresponding equivale nt circui t is shown in Fig. 5-2b, and
Fig . 5-3 Phasor dia gr am of a
the corresponding ph asor dia tra nsformer with it s secondary
short-circuit ed (10 ;:::: 0)
gram in Fig. 5-3. As is seen
from the ph asor diagram , the
short-circuit voltage V 1 = Zse 11 is the hypothenuse of a
triangle whose legs are the active vol tage R sel 1 an d
the reactive voltage jX se 1 1
Th e right-angled voltage (or impedance) tri ang le drawn
for t he sh or t-circuit condition is referred t o as the shortcircuit triangle, and the angle

<Pse = arctan (Xse/R se)

is called the short-circuit angle. On setting also [see Eq .

we will obtain still simpler expressions for the magnetizing

current and emf

~1 = -:j~ = V1/~se


V 1/ 2Z o

I o,oc/ 2

= zoi 0 , oe/2 = 111./2



Part One. Transform ers

H ence it follo ws t hat the referred secon dary current on short

circui t does no t diff er from the primary current, which is
also t rue of the rated-lo ad condition; the magnetizing cur ren t is ab out h alf as lar ge as the no-load (open-circuit) cur rent with the same prim ar y v olt age appli ed; the mutual
emf is ab out h alf as large as t he open-circuit primary voltage
or emf .
If the secondary is sh ort-circ uited wi th t he r ated primary
voltage applie d, VI = VI. TI, t he n t he tra nsient s (to be
dis cussed in Sec. 13-2) will giv e rise to short-circui t primary
and secondary current s [see Eqs . (5-6) an d (5-7)1 dangerous
to the transformer . On a per-unit basis [see Eq. (3-33)1
these currents are
1,~ 2

= V*l. TIl I Z*. sc I = 7 to 16

that is, they are 7 tuIf t imes as he av y as the ra te d currents
in the win dings .
If such currents were all owed to exist for a lon g t ime, the
resultan t t emperature rise would impair the electrical and
mechani cal stre ngth of the insulation . For this rea son,
a short-circuit test is con ducted at a reduced primar y voltage
wh ose va lue is chosen suc h that t he currents in the windings
coul d not exceed their ra ted va lues .
On a per-uni t ba sis , this voltage should not exceed

< v'n ,sc = I Z'!:sc 11,':1, R = I Z*, sc I = 0.06 to 0.14

A likely set -u p for a short-circuit test is sh own in Fig. 5-1b.
As in the open- cir cui t t est , it does not require any bulky
load resistors or a large te st voltage source. (Wi th t he shortcircuit current ad jus ted t o the rated value, the power
r ating of the test source is not over 0.06 t o 0.14 of t he power
rating of the tr ansformer under t est.) The voltage is gradually raised from zero t o anywhere from 0.06 to 0.14 of the
rated primary voltage. While holding the frequency at its
rated value, I = I r a t erl' r eadings are taken of the sam e qu antities as in the open-circuit te st, namely the prim ar y current
II and the power drawn by t he transformer, P scUsing t hese readings, I I' P sc an d cos cvsc ar e pl ot t ed as
functions of the phase voltage VI' and t he plo ts thus obtained
are us ed to det ermine graphically VI SC , P c and cos <Psc
at the rated primary current , II = I I n . For a three-phase
transformer, the plots are constructed for the av er age ph ase
IIi and the av er age
VI' and the power

Ch. 5 Measu rement of Transform er Quantities


fac tor is found from the average values of 11 and VI as


P se/3TTl Il

crse =

W ith the cur rent m ai nt ained at its r ated value, the shortcir cuit te st yields the following transformer quantities.
1. The short-circuit impedance from Eq. (5-7) as

I Z 5e I = VI , selIl, R
its resistive component

I z., I cos

R se = P se/3IT. R


and its reactive component

X se

= V I Z~e I -

R~e =

I z., I si n


The resistive component is the sum of the winding resistan ces, R se = R 1 R ;. During a short-circuit test, it
is important to note th e winding temperature e at which
R se is measured. Th e me asured value of R ~e is then adjusted
t o a temper ature of 75C:


R se. 7 5 = tt.; [1 + 0.004 (75 The reactive component of the short-circuit impedance
is the sum of the leakage inductances, X se = Xl
whi ch , as has been explained in Chap. 3, are independent
of the current s tra versing t he respective windings. For the
sa me reason , X r e is in dep en dent of the current at which
it is measured .
The short-circuit impedance and power factor are likewise
adjusted to a temperature of 75C:

I Z se, 7 5 I =

11 R ;e, + X~e

cos crse , 75 = R se , 75 / 1 z.;


2. The short-circuit loss Pse . At II = 11, R, it does not

practically differ from the copper losses in the primary and
second ar y windings carry ing r ated currents

Peu. R =

P e ll . r , R


sn 1


10 R

+ Peu,
+ 3R ' I '2

because the copper loss

loss, P r;gre . W .

2. R

= 3R1I~, R

2 2' R =


3R sc 1 1.2

+ 3R

2I:. R

Illany times the short-circ uit yQn:l


Part One. Transformers

With VI, se equal to 0.06 to 0:14 of VI, R, the short-circuit

primary emf is
E I ; Be = VI , se/2 = 0.03V I, R to 0.07V I, H
The short-circuit flux and induction in the core, which are
proportional to E I , sc- amount to anywhere between 0.03
and 0.07 of their open-circuit values . The short-circuit core
loss, which is proportional to the magnetic flux density
squared, ranges between 0.9 X 10-3 and 5 X 10-3 of the
core loss under rated con ditions, and between 2 X '10-4
and 12 X 10-4 of the copper loss also under rated conditions.
3. The impedance voltage. It is defined as the voltage
that must be applied to one of the wind ings, with the other
short-circuited, so as to circulate rated current at a tem perature of 75C, with the windings connected as for rated
voltage operat ion. If this voltage is ap plied to the pr imary
winding, the impedance voltage expressed in abso lute units is

VI, se = I z.; 75 I II , R
Usually, the impedance voltage is expressed in per unit
or per cent of the rated voltage of the winding in which the
voltage is measured

se = V*l, sc = VI;se/ VI; R = Z",, se

(VI, se/ VI, R) X 100%
Similarly , the resistive component of the impedance voltage is given by
Va = R se , 7511, R/VI R = R*, se = V se COS CPsc
and it s reactive component is given by
u; = XseI I, R/V I , R = X 'I', se = V se sin CPse
(see the ph asor diagr am in Fig. 5-3).

V se

Tra nsformer Performa nce on Load


Simplified Transformer Equations

and Equivalent Circuit for 1 1 1 0

In service, the load on a transformer is varying all the time.

As a result of variations in the load impedance Z, the secondary current may vary from zero to its rated va lue, and it~
phase relative to the secondary yolta~e also varies ,


CII. 6 Tr an s form er Perform an ce on Load

As h as been expla ine d in Sec . 3-7, variati ons in the secondar y curren t are accom panied by nearly proportionate variat ions in the primary cur rent , and th is le ads to slight variations in the.magneti c flux . This ch apter will deal with the
effects that variations in the
secondary current m ay ha ve
on t he secon dary voltage an d
efficiency of a t ransformer .
", -VI Analysis will be carried out
for the most frequently encountered load cond itions, namely VI, R = constant and
II 1 0 ,
Fig . 6-1 Simpli fie d equivalen t
When 11 10, we may, as
circuit of a transformer with
II ~ 1 0
in the case of a short-circuit ,
set 1 0 =0 and IZo 1= CXJ .
On this assumption , the primary current [see Eq . (3-20)]
does not differ from the secondary current referred to the
primary 11 = - I;, and the voltage equations for the
prim ary and secondary wi ndin gs , Eqs. (3-13) and (3-19),
may be combined in to a single equation


+ Zscll
R sc + jXsc is



where Zsc = ZI
Z~ =
the short-circuit
impedance of t he transform er . Therefor e, the equivalent
circuit in Fig. 3-6 m ay be simplified by removing the arm
carrying the magnetizing current, and the sum of im pedances ZI
Z~ m ay be replaced by Zsc. The simplified
equiva lent circui t an swering Eq. (6-1) appears in Fig. 6-1.


Transformer Voltage Regulation

Graphically, th e depen dence of V 2 on 1 2 , with the power

factor cos CPl\ an d VI, R held constant, can be shown by an
external characteristic. Plotted in arbitrary units, it takes
the form shown in Fig. 6-2. The manner in whi ch V 2
var ies with 1 2 depen ds on the char act er of load . If the load
is resis tive-inductive ((P2 > 0), V 2 decreases as 1 2 increases.
If the load is predominantly capa citive ((P2 ~ - n /2), t he
on-load secondary voltage may exceed its rated no-lo ad
va lue ,


Part One. Transformers

A measure of how much

changes in 1 2 will cause V 2 to
change is given by voltage
change defined as
L1V2 = V 2 R



with VI = VI, R held constant.

I t is usually expressed . in
per-unit and known as voltage

l '

Fig. 6-2 External characteristics, V 2 = f (1 2 ) , of a transformer with VI held constant: solid

line, resistive-inductive load,
CJl2 = const > 0; dashed line,
resistive-capacitive load, CJl2
= const < 0

L1v= L1 V 21V2 R


(VI, R -

R -

= L1 V' lVI, R




R -

= 1 - V'I'2



If the per-unit voltage regulation is known, the per-unit

secondary voltage can be
found from

Fig. 6-3 Simp lined phasor diagran, of transformer voltage

Te~ulation~ with I~ ~ I Q

The voltage regulation equation can be derived from the

voltage regulation diagram
in Fig. 6-3. (For convenience,
the voltage drop phasors
R:I:.scI I and X* .scII are shown
on an exaggerated scale.)
Construction of the voltage
regulation diagram begins
at point 2 which is the tip
of the - V~ ph asor . The - V~
ph asor is drawn through this
point in an arbit ary direction,
and I I = I ~ s then drawn
to make an ngle ep~.with the


Ch. 6 Transformer Performance on Load

- V~ phasor. Now point 2 is used as the origin fo: the phasor

representing the resistive voltage drop R scl1 and the
reactive voltage drop jX s
Point 1 occurs at the tip of
the VI phasor. The value of - V~ is found at the intersection
of the - V~ phasor and the circle with point 1 as centre a~d
with VI. R as radius. At the same time, the direction of VI
and the angle a between VI and - V~ are found.
The diagram in Fig. 6-3 has been plotted in per-unit and
its components are given [see Eq. (5-10)1 by the following



V*l. R = VI, R/V1, R = 1

1*1 = II/II, R = 1*2 = 1 2/12, R = I~/I1, R = ~

V~/V1. R

R*. SCI *l = R sc (II. R/V1, R) 1*1 = Va~

X*, scI*l = Xsc (II. R/V1, R) 1*1 = vr~
Let us write the voltage regulation as the difference between V*l, Rand V*2:
"' /),v = V*l,R - V*2 = 1 - V*2
Referring to the voltage regulation diagram

V*2 = V~2 = V'H, R cos a - c = cos a - c

In turn, because the angle a is negligibly small
cos a= V 1- sin 2 a ~ 1- (sin 2 a )/2 = 1-d2/2
On expressing the segments c and d as the sum and difference of projections of R*, SC!,H and X'i<, SC!,i<l on the V~2
direction and on a direction at right angles to it

+ X*. sc sin CP2) 1'1<1

c = (R'I<, SC cos CP2

= (Va cos CP2
d = (X'I<,SC cos CP2 = (vr cos CP2 - Va

sin CP2) ~
R*. sc sin CP2)
sin CP2) ~


we obtain an expression for the voltage regulation

/),v = c d2/2 = (va cos CP2 o; sin CP2) ~

+ (vr cos qJ2 -~a sin qJ2)2 ~2



Part One. Transformers

As is seen, the secondary voltage regulation depends substantially on the load phase angle (P2' A plot of !'!..v as a function of cr2 with ~ = 1 for a transformer with Vse = 0.1.
Va = 0.04 and u; = 0.0918 is shown ill Fig. 6-4. The dashed
line gives the same dependence on neglecting the second




tf I


Fig. 6-4 Voltage regulation, S, as a function of (jl2, with 1 2 = 1 2 , R

ana ~ = 1:
solid line, by Eq. (6-3); and dash ed line , with the second term in the
equation ignored

term in the equation. Because the inclusion of the second

term affects the final result but little, an approximate
equation, convenient for analysis,

(va cos cr2

u; sin cr2) ~
~vse cos ((Pse - cr2)
is used in many cases (especially where Vse is low).
As follows from Eq. (6-4) the voltage regulation is a
maximum, !'!..v = Vse, when cr2 = crse, because cos ((Pse
- cr2) = 1. Conversely, the voltage regulation is a mini0
mum, !'!..v = 0, when (Pse - (P2, 0 = 90
cr2, 0
= _(90 crse), because then cos (crse - (P2, 0) = 0 (see
Fig. 6-4). With some other load phase angles, the shortcircuit triangle takes up the characteristic positions shown
in Fig. 6-4, such tha\
!'!..v = Va
for CP2 = 0
!'!..v = +vr ~ = +90


Ch, 6 Transformer Performance on toad

Becau se the second term in Eq. (6-3) is small, the dependence of voltage regulation on the relative secondary cur rent ~ , with qJ2 held cons tan 1., is practi cally linear.

Variations in Transformer Efficiency

on Load

Electric energy should preferably be transformed with as

low relative losses as practicable or, which is the same, with
as high an efficiency as can be achieved. Here, t he efficiency of a transformer is defined as the ratio between the
active power delivered to the line on the secondary side
and the active power drawn from the supply line on the
primary side
11 = P 21Pl = m2V212 cos qJ21mlVlIl cos



The primary active power may be written

PI = P 2

+ P eor e + P eu. + Pc

We shall limit ourselves to the operation of a loaded

transformer with the rated primary volt age, VI . R' , held
constant. We shall make the same assumptions as in Sec. 6-2.
That is, we shall deem II ~ 1 0 , 1 2 = I~, and I Zo I = 00.
We will also neglect the difference in core loss between
operation on load and at no-load and assume that
P eor e

P eore, oe


where Po is the no-load loss with the rated primary voltage

appli ed. Then the copper losses may be expressed in terms
of the short-circuit loss P se at the rated primary current:
Pee ,

+ P eu.

= I~Rl

+ I~2R ~

= IiR se

I~ .RR se (lIllI, R)2 = Pse~2

The secondary active power is given by*

P 2 = m2T{2Iz cos (P2 = mlVl. RI~ cos qJ2
_ _ _ _ ml
RI I R (I~lIl, R) cos qJ2 = SR~ cos eP2


* In setting V~ = ~R' we negl ect the effect of voltage regulation

on the secondary active "power .


Part One. Transformers

Substituting the above ex pressions in E q. (6-5) gives us

the dependence of the effi ciency on ~:

P l-(P COl'e -/-PeU .I -/-P cU. 2)



BSR cos CP2+ PO+ B2P sC

(6 6)

The effect of the secondary current on t he secondary voltage may be accounted for as follows :
V~ =

VI, R (1 - L1v)
and its effect on the iron loss thus:
Pc ore = Po (EIIV I, R) 2 = Po (1 - L1v)
where, with sufficient accuracy ,
E I = VI, R (1 - L1vI2)
Accordingly, the equation for the secondary active power
may be re-written as
P 2 = m 2 V 2I 2 cos CJl2 = mi V~I~ cos CJl2
= SR~ (1 - L1v) cos CJl2
Then, the effi ciency equation ma y be refined as

Po (1- ~ v) +B 2Ps c

r3(1- ~v)SRCOS (P2+Po(1 -~v) +B2Psc

(6 ~)

Equation (6-7) holds for the entire range of changes in

secondary currents. On both open- and short-circuit

P2 =

~ (1 -

L1v) S R cos CJl2 = 0

and the efficiency reduces t o zero . This can be pr oved form ally from Eq. (6-7) , recalling th at at no-lo ad ~ = 0,
whereas on short-circuit, L1v = 1 - V 2* = 1, because
V 2 :1: = O.
Although approximate, Eq. (6-6) derived for ' L1v = 0
is sufficiently accurate for the relat i ve secon dary current
varying from ~ = 0 t o ~ ~ 1. Let us fi nd the value of ~
at whi ch the effi ciency is a maximum . E quating the deri vative df]/d~ t o zero and simplifying t he equat ion, we obt ain


~ ~axPsc =

~ ll1ax =

f -

-I- Peu.

1- PolP sc





7 Tap Changing

This implies that the efficiency of a transformer is a maximum when the load is such that the no-load core loss at
rated primary voltage, Po,
is equal to the copper loss,
Peu. 1
P eu. 2'
cos <jJ,-O.8
In presenl-day power trans'/
formers, PolP sc ranges from
0.2 to 0.25, and the effic iency
is a maximum at ~ = 0.45
to 0.5 .
Efficiency curves for a
three-phase transform13
CP2 = 1 and cos CP2
= 0.8 are shown in Fig. 6-5.
As is seen, Po = 2.45 kW
Fig. 6-5 Effi ciency, T], as a
and P sc = 12.2 kW. The effifunct ion of the relative load
ciency is a maximum at

~ rn a x =

V 2.45/12 .2


In the range from O .4~max to 2.5~rnax, the efficiency falls

off insignifi cantly. Such variations in efficiency are typical
of all power transformers.

Tap Changing


Off-Load Tap Changing

As follows from the analysis given in Sec . 6-2, in the worst

case (when the load phase angle CP2 is equal to the phase
angle on a short circuit, CPsc), the per-unit voltage regulation
may be anywhere between 0.06 and 0:14 . Th is is far more
than is permitted by relevant service codes. To maintain
the secondary voltage constant against such variations,
tappings on the coils are brought out to terminals so that
the number of turns can be changed.
This tap-changing can be effected on either the primary
or the secondary side. In transformers operating at a fixed
pri mary voltage, this is done by changing the number of
turns on the secondary side, while holding the primary
turns unchanged . With this arrangement, t he magnetic
flux, the core l oss, and the magnetizing current (whi ch is
a function of the ratio V1/wl) remain practically constant.
8- 0160

Part One. .Transformers

in transformers operated at constant load (or, which is

the same, at a constant secondary current) and a varying
primary voltage, VI' it is preferable to change taps on the
primary side so as to maintain the ratio VI/WI nearly
All power transformers have
tappings on the primary
or secondary coils which permit voltage adjustment within
+5 %. Low- and mediumpower transformers (Fig. 7-1)
usually have three taps per
phase (+5 %,0, and -5 % variations in the turns ratio),
whereas transformers of higher
power ratings have five taps

(+5%, +2.5%, 0, -2.5%,

and -5 % variations in the
turns ratio). Tap stepping operations are performed by contact switches, usually called
tap changers. Tap changers can
be made simple and inexpensive if taps are stepped with the transformer out of circuit
and the taps are made at the neutral point of a three-phase
star-connected winding (see Fig. 7-1). This avoids a short
circuit between adjacent taps or breaking a live circuit
during a transition. The operating handle of the tap-changer
is passed outside through the tank side.
Fig. 7-1 Tap-changing by a
1-transformer winding; 2-tapchanging switch


On-Load Tap Changing

Voltage adjustment can be made far more accurate and

automated if tap changing is done with the transformer left
on load, without breaking the circuit. This, of course, calls
for a more sophisticated tap-changing arrangement, notably
one incorporating what is known as a transition impedance.
Impedance in the form of either resistors or iron-core inductors is introduced to limit the circulating current between
the two tappings.
Most frequently, resistor transition is used ~o r on-load
g" Ior
tap change". The arrangement of such a tap



eh. 7 Tap Chan gin g

one ph ase is sh own in Fi g. 7-2. This is seen to be a combination of a fa st-acting di verl er switch D S , an ev en t ap selector
TS l' an d an odd tap selec to r T S 2 ' The di ver ter switch and
the transition resistors, R 1
and R 2 , are usually in stalled
in a sep ar ate oil-filled t ank.
The di verter switch is designed t o carry t he current usually developed when the two
taps ar e bridged . The tap
selectors may be moved from
tap to tap onl y when their
circuits are de-en ergized . Figure 7-2 shows t he diverter
swi t ch an d t he even tap sel ector in the position when the
T 2 tap is brought in circuit.
To move to the next, T 3 ' tap,
Fig. 7-2 On-load ta p changer
t he odd t ap select or should
with cur rent-li miting resistors
first be moved t o that tap,
and the diverter switch ma y
then be rotated clockwise. The ensuing sequence of events
is as follows: con tacts . 1 an d 2 break , contacts 1 and 3
make, contac ts 1 and 3 br eak, and contacts 3 and 4 make.
When fully automated , a tap stepping operation is completed in a m atter of a spli t secon d.

Fig . 7-3 On-l oad t ap- changer with a transition i nducto r


The arr angement of a t ap

one phase) whi ch
uses the t ra nsit ion impedance in the Iorrn of an iron-cored
in ductor is shown in Fig. 7-3 . In addi; ion to a transition
cent re-t ap inductor (or rea ctor), L , whi ch is wound in two
hal ves , 1 an d 2, put on a comm on no-gap core, the arrangement includes two t ap sel ectors , TS 1 and TS 2 ' whi ch can


Pa rt One. Transformers

move fro m tap to tap after their circuits have been de-energized, and two on-off switches , 8 1 and 8 2 , to de-energize
the respective tap-selector circuits.
The tap selectors and the centre-tap inductor are located
in the transformer tank, and the on-off switches are enclosed
in a separate tank mounted on the transformer. In Fig. 7-3a,
the load current is shown passing from tap T4. through the
halves of the inductor in opposition, and hence noninductiv ely (wi t hout magnetizing the inductor core). Therefore, t he
inductor presents to the load current onl y a small resistance,
while it s reactance may be igno red . Transition from tap T 4.
10 , say, tap T 3 may be visualized as consisting of a sequence
of seven steps li st ed in Table 7-1. The most significant steps
Table 7.1 Tap -to -Tap Transition Steps






T8 2





Figur e No .


7-3b (dashed)

are illustrated in Fig . 7-3. In Fig. 7-3b, one of the two tap
select ors ha s opened , and the load current is carried through
one half of the inductor induct ively (setting up a magnetic
flux ar ound it). However, the induct or is designed so that
the instantaneous reactive vol tage drop during this step
ha s but an insignificant effect on the secondary terminal
voltage of the transformer. In Fig. t:3~ the inductor is
sh own bridging the two adjacent tappings,"T 4. and T 3. The
load current is shared equally between the~two tappings
and passes noninductively in opposition through the ha lves
of the inductor. The tap step voltage is applied to the whole
of the inductor wind ing and the circulating current, I e'
is limited by the total impedance of the inductor whose fie ld
is now directed aiding to the mmf due to the circulating
current (shown by the dashed lines in the fig ure ).


Ch. 8 Calculation of Transformer Parameters

Calculation of Tra nsformer

Pa rameters
No-load (Op en-Circuit) Current
and M utual Impedance


In Sec. 2-6 it has been shown that the reactive component

of the no-load current, Ion can be deduced from the parameters of the magnetic circuit. It is, however, simpler and
more convenient to determine it from the reactive power
required to magnetize the transformer. The reactive power
may be expressed in terms of either the mutual emf and
the reactive component of
the no-load current
Qo = s.r;
(for a single-phase transformer)
lsap or the core flux. Let us do
this with reference to Fig. 8-1.
In terms of the peak flux,
E l is given by


Fig. 8-1 To calculation of reactive currents required to magnetize the core, [or core, and
the gaps, [or, ga p


= 2'Jf,fw/Pm/ V2


In accord with Eqs. (2-13) and (2-14), the reactive component of no-load current can be written as

lor = lor, core

lor, gap
The first term on the right-hand side sustains the magnetic
potential drop in the core
lor, core = H mlcore/ 112 Wi
and the second term sustains the. magnetic potential drop
in the air gap
lor, gap = B ml gap /;l/ 2 Wi~to
The total reactive power is the sum of the reactive powers
required to set up the flux in the core and the gap
Qo = Ell or = Ell or, core


'+ Acoreqgap

Ell Or,



Part One. Transformers

Ell or core/mcore = niBmH n/r is the specific

magnetizing power of the core
qg ap = nIBinlgap/f.lo is the specific magnetizing power
of the air gap
m cor c = mass of the core
A core = cross -sectional area of the core
r = density of the core material
The values of the specific magnetizing power as a function
of induction (magnetic flux density) for imbricated-joint
cores are given in [131. The reactive power required to
magnetize a core of any design is given by



qco r e =


+ qyo] lemyoke + n ga p , l egqgap, legAl eg

+ n g a p , y ok eq gap , yokeAyoke


where ml eg and myo]w are the mass of the legs and yokes,
A 1e g and A y ok e are the cross-sectional areas of the leg and
yoke, and ql e g' qYol,e' q ga p, 1 e g and q gap, yol!C are the .specific magnetizing powers of the legs, yokes, leg gaps, and
yoke gaps .
In a single-phase core-type magnetic circuit, n g a p , leg = 2
and n ga p, y oke = 2. In a three-phase core-type magnetic
circuit, n g a p , leg = 3, and n g a p, y oke = 4.
The active power, equal to the core or no-load loss, is
deduced from the specific loss for legs, PI e g' and yokes,
P yoke' which are given in [13J

This power is ordinarily calculated for the rated primary

voltage, VI, R = E I , R, only. Once it is found, it is an easy
matter to determine the no-load current components (see
Sec. 2-8):



loa =Po ~

the no-load current

10 = 11IBn

+ 18;'

and the components of the mutual impedance

Ch. 8 Calculation of Transformer Parameters

"* 8-2


Short-Circuit Impedance

On a short circuit (see Sec. 5-2), the primary mmf, i1w1,

and the secondary mmf i 2 w 2 , balance each other almost
completely. Therefore, without
running into a serious errol',
wemaydeem that i1w1= - i 2w2,
and that, on a short circuit,
only the leakage flux exists,
whereas the mutual flux is
non-existent, because
iOw1 = i1Wl
i 2w2 = 0
The flux pattern applicable
to this case, with the windings
arranged coaxially, is shown
in Fig. 8-2. The magnetic
field intensity H within and
between the windings acts
along the leg axis. With sufficient accuracy, the magnetic
field may be taken as being
symmetrical about the leg
axis. Therefore, the value
of H remains practically the
same within a distance D/2
of the leg axis and along the
coil height h. The magnetic
intensity distribution along
the radial coordinate x reckoned from the inside surface of
the coil area is shown in the
same figure.
By Ampere's circuital law,
the magnetic field intensity
is a maximum between the
windings (a 1 < x < a 1
a1 2) ,
where the magnetic lines of
force link all of the]JrIiilary
Fig. 8-2 Leakage flux in a
transformer on a short-circuit
(i1lV 1 = - i 2 lV 2 )
H = H m ~ i1w1/h

In a first approximation, the magnetic potential drop in

a core material with m-, core = CXJ and H core = 0 may be


Part One. Tra nsform ers

neglected. W ithin the wind ings, t he magnetic fie ld intensity

varies linearly from zero to H m- For ex ample, when x is
anywhere from zero t o aI' t he magnet ic lines of force link
with a current i1wrx/a r, and the magnetic fie ld int ensit y is

H = i1wrx/ha 1 = H mX/al
The energy associated with t he leak age flux established
by two magnetically coupled windings may be expressed
in terms of the inductances of those windings
Recalling that


+ L~i~2/2 + iri~L~2

we obtain

= (L ru

+ L;u) i~/2



where L sc is the short-circuit inductance.

The same energy may be expressed in terms of the sp ecific
energy of the magnetic fie ld

= HB/2 = flOH2/2

In determining the energy, the integral may only be taken

over the volume V = nD mean (a1
a r2
a 2) 12, occupied
by the windings, which encloses the bu lk of the le akage

W = .\ w dV

= (flo/2) .\ H2 dV

Because H remains the same all the wa y round the circle

with the diameter D = Dr
x and along the height 12,
the elementary volume is .


= nDh dx = n (Dr

+ x )h dx

With an accuracy sufficient for engineering purposes,

the diameter D = Dr
:r may be replaced with a mean

Ch. 9 Transformer Quantities vs. Dimensions


After this simplification,


W = (l1-oh:rtDmean/2)


HZ dx = (11-0/2) nDmeanhauH;"
.: (8-4)

a z)/3.
where au = a l 2
Equating (8-3) and (8-4) yields an expression which ' connects the short-circuit inductance to the size and winding
data of a transformer
L S0 = nl1-oDmeanwiaukRlh
k R = 1 - (al .+ al2
is the Rogovsky coefficient (after its originator). It minimizes the error in calculations due to the assumptions made.
The short-circuit inductive reactance is given by
X se = (2n zfl1-o D meanauw~) k F l h ( 8 - 6 )
The resistive component of the short-circuit impedance
is calculated as the sum of the referred resistances of the
n.; = n, R~ = n, tt, (w l/wz)2
where R l = P7.'inDmeanWlks/Awl is the resistance of the
R z = P75nDmeanwzkslAwz is the resistance of the
AWl' A wz = cross-sectional area of the primary or secondary turns, respectively
P75 = resistivity of the wire at 75C after [13]
k, = 1.05 to 1.15 is the series-loss coefficient.

Relationship Between Transformer

Quantities and Dimensions


Variations in the Voltage,

Current, Power and Mass
of a Transformer with Size

Suppose we have a range of geometrically similar transformers. Two transformers out of this hypothetical range
are shown in Fig. 9-1. Any dimension of any transformer


Part One. Transformers

in this range differs by a factor of k from the like dimension

of any other transformer in the range .
Taking any linear dimension, say, the height, l, of the
core as the base or reference dimension, we may deem that
all the other dim eilsions of the
tr ansformers in a given range
are proportional to it. For
example, the mean turn diameter, D mean , is about equal
to l. The cross-sectional area
of any element of a transformer is proportional t o the
square of the base dimension ,
A ~ P . By the same token,
the v olume of any element is
proportional to the base dimension cubed, V ~. P .
Now let us see how t he
rated electromagnetic quantiFig. 9-1 Geometrically similar
ties of a transformer are reI asingle-phase transformers
ted t o it s size. Suppose that
all the transformers in a given
range are fabricated of the same materials and that the
magnetic flux density B in the core, the current density J
in the wind ings, and frequency remain always constant .
1. Neglecting the difference between terminal voltage
and generated emf, we get

VI. R ~ E 1 ,

= (2n/ V 2) !wiBA Jeg ~ w:rAJeg """ w'l.ZZ


That is, the primary voltage is proportional to the number

of turns in the primary winding and the square of the base
dimension .
2. Assuming that as l is varied , the total cross-sectional
area of conductors in a wind ing r em ains proportional to l2,
we get

That is , t he primary current

of the base dimension . (Here,
area of the conductors in the
3. The total (or apparent)


= VI, nIl, R

is proportional to the square

Al is the total cross-sectional

power of a transformer

l2WI (l2/ w1)

= l4



Gh. 9 Transformer Quantities vs. Dimensions

is proportional to the base dimension raised to the fourth

power . Importantly, on the assumptions made, the power of
the transformer is independent of the number of turns.
It. The mass of the transformers in a range fabricate d
of the same materials

m = ~ I'll "-' [ 3
is proportional to the base dimension cubed.
The mass per unit power
mJS "-' [3/ [4 "-' ill ("",1/S1/4 )

is inversely proportional to the base dimension. (The mass

per unit power decreases as power r ating goes up.)

Transformer Losses and Parameter s

as Functions of Size

1. The total loss of power in a transfor mer is t he sum of

the core loss and t he copper loss.
The core loss
is proportional to the mass of the core elements, mcore,
because the specific core loss in similar elements remains
unchanged as the physica l dimensions are va ried .
The copper loss may be expressed in te rms of the winding
112 = nDmeanA 2
and als o the current density J and the resistiv ity p

Pc = Peu. 1

+ Pc.

pJ2V 2
If the materials remain the same and the current density
is he ld constant, the copper loss is prop or t ional to the base
dimension cubed

P eu = pJ2 (VI
, Thus, t he transform er loss
P col'e

+ V 2) "-' [3

+ r-; "-' za

is proportional to the b ase li near dimension cub ed . :




Part One. Transformers

The spe cific loss (the t ot al loss divided by to tal power)

(P eu

Peore)/S '" l31l 4


ill '" 1/S1 / 4


is inversely proportional to the base linear dimension or

the fourth-power root of the total power. (In high-power
transformers, the spe cific loss is lower.)
The loss per unit of cooling ar ea, A e oo1

(P eu
P eore)/A eool '" ZSll2 = l
is proportional to the ba se linear dim ension and increases
with increasi ng power rating. This is t he reason why highpower transformers must be pro vi ded with a well-developed
cooling are a in the form of ducts in t he core and windings.
2. The short-circuit inductive react an ce (see Sec. 8-2)
X se '" w~Dmeanaa/h '" w~ l


is proportional to the number of turns squared and the base

linear dimensions.
The short-circuit resistanc e
R se = Pe ulI;. R


l3/(l2/W1)2 '" w~/l


is proportion al t o t he number of t urns squared and inversely

proportional t o the base linear dimension.
The reactive and resistive components of the impedance
(short-circuit) voltage are given by


R*ae =



Xsel l R/V1 , R '" (Wil/Wll2) (l2/WI) '" l

= Raell R/V1 R '" (wi/l) (l2/WI) (1/w ll 2) '" 1fl

The short-circuit tangent is

tan lVse

= X *aeIR*ae = vJ; '" 1fl2


In other words, as a t ransformer gr ows in size , its v; rises

and its va falls. This checks well with practice.
3. The reactive and active components of t he no-load
current (see Sec. 2-6)
lor '"

~ Hz. dl/wi '" llur,

loa = Peore/Vj.

R '"

Z3/w jl2 ,....,. 1/wj


are proportional to the base linear dimension and inversely

proportional t o the number of turns.

Ch. 10 Multiwinding Transformers. Autotransformers


The relative no-load current (or the relative magnetizing



1 0 V I , Rll l , RV I , R
'" (l/w l ) (w I/Z2 ) = ill

= lolli, n


is inversely proportional to the base linear dimension.

In commercially available transformers, geometric similarity is never complete, nor can Band J be held constant.
Nevertheless, the relationships set forth above are true, at
least qualitatively.
As follows from the foregoing, it is advantageous to use
transformers with higher power ratings, because they take
less materials per unit power, need lower reactive power,
and dissipate less heat.


Multiwinding Transformers.


Multiwinding Transformers

(i) Three-Wlndlng Transformers

In a multiwinding transformer, the core carries.more than
two electrically isolated windings.
Power systems mostly use three-winding transformers to
couple electric systems or networks operating at three different voltages, VI' V 2 , and V 3 Three-winding transformers may be single-phase (Fig. 10-1) and three-phase, with
their windings connected Y n/Y nl~-0-11 (Fig. 10-2) and
Y nlM ~-11-11.
A three-winding transformer may have either one primary (1) and two secondaries (2 and 3), or two primaries
(1 and 2) and one secondary (3). Our discussion will be limited to transformers having one primary and two secondaries.
A three-winding transformer does the job of two twowinding transformers one of which connects network (or
system) 1 to network 2, and the other, network 1 to network 3. Economically, a three-winding transformer is more
attractive than two two-winding units. Among other things,


Part One . Tra nsformers

it is less ex pensive to make and takes up less space at a

substat ion It can transf er power not onl y from the primary


Fig. 10-1 Sin gle-phase, three-winding transformer:

i-primary winding ; 2-secondary win ding ; 3- tertiary win ding; 4she ll-and-c ore-type m agnetic circuit

network (1) to any of the secondary networks (2 or 3), but

also directly (by a single transformation step) from one of

Fig . 10-2 Y n/Y nl ~ -O- 11 t hree-phase , t hree-win ding t ransformer :

i -LV three-ph ase win din g; 2-MV t hree- phase win din g; 3- HV
t hree-phase win ding; 4-core-type magne tic circuit

the secondary networks to t he ot her (say , from network 2

t o network 3). With two t wo-win ding t ransformers, such
power tra nsfer necessitat es two transfor mation st epsfirst from network 2 t o network 1, then fro m network 1

eh. 10 Multiwinding Transformers. Autotransformers


to network 3. Accordingly, the losses are about twice as

On the demerit side, a three-winding transformer is less
reliable. Should any of its windings be damaged, the entire
unit must be removed from service. With two two-winding
transformers, damage to any of them leaves the other unaffected .
The electromagnetic processes in a three-winding transformer may be described by analogy with a two-winding
unit (see Chapters 2 and 3) . As a preliminary step, however,
its secondary and tertiary quantities must be referred to
the primary side, with their values multiplied by the respective turns ratio or its square:
I~ =


(WZ/w I)

I~ = Is X (Ws/wI )

= V z X (wi/W Z)
= V s X (W I/W 3)
I = I z, I X (wI/WZ) Z
I = I z, I X (WI/W S)2

I Z~
I Z~

The mutual flux is set up by a magnetizing current 1 0

which is given by the current equation


+ I~ + I;

= 10


and the mutual emf is given by







Zo = R o
jX o is t he mutu al impedan ce.
leakage flux is established by a balanced (or symmeset of currents, j~ , j~, and I~, where i; = i; - i;
leakage emf in each winding is

E. UI =

-jXI~~ -jXI~


E~2 =



-j X ;I~

where X l' X~, and X; are the equivalent leakage reactances

of the windings, found wi th allowance for the effect of
currents in th e other ' windings.


Part One. Transformer s

Formally , the vol t age equations for the t hree wind in gs

are written as for a two-wind in g t ran sform er [see Eq s. (3-13)
and (3-19)1:

+ RIll

-E~ = - ir~ + E~2 - RJ~ = - V~ + Z~ (-j~)
-E~ = -i'; + E~3 - R/; = - V; + i ; (- j ;)

-E I - E cn


Z~ =
Z~ =

-E 1



+ jX
R ~ + jX~
R~ + jX~

Equations (to-1) through (10-3) apply t o t he equ i valent

circuit in Fig. 10-3.
The mutual impedance Z~ is foun d by calculation or
experiment in exactly t he sam e way as for a two-winding tra nsformer (see Sees. 5-1 and 8-1). The impedances Zl' Z~ and


Fig. 10-3 Equivalent circuit of a three-winding transform er


are exp ressed in te rms of the sh or t-circuit impedances

an d ZSC23 ' as determined by a short-circuit t est,
using t he tes t set-u p shown) in Fi g. 10-4. ZSCl 2 = R SCl 2
jX s c l 2 is found with the te rti ary winding open-circuited.
Z SCl3 = R SCl 3
jX s c l 3 is found with th e second ary winding open- circuited. Finally Z SC23 = R SC 2 3
jX s c 2 3 is
. found wi th t he secon da ry energized a nd t he prim ar y opencir cuited , and is referre d t o the prim ar y by t he equation
Z SCI 2 ' Z SCl3

Z~C 2 3

It is to be noted that

Z SC2 3



Ch. 10 Multiwinding Trans forme rs. Autotransformers

Solving the above equations for ZI '


Z~ ,



we get

= (ZSCI2 + ZSC13 - Z~ C 23) /2

= (ZSC12 + Z~ C23 - ZSCI3) /2
= (ZSCI3 + Z~ C2 3 - ZscI2)/2


The resisti ve components of th e ab ove imped an ces are the

resi stances R I , R ~ and R ; of th e respecti ve windings (see






Fig. 10-4 Short-cir cuit tes t on a single-phas e, three-winding transformer

Sec. 8-2), whereas their reactive components have the meaning of the equivalent leakage inductive re actances of the
x, = (X SCI2 + ~XS CI3~- X~C23)/2
X~ = (X SCI2 + X~C23 - X SCI3)/2
X; = (X SCI3 + X~C23 - X SCI2)/2
The secondary and tertiary voltages of a loaded threewinding transformer may be found analytically, using
Eqs. (10-1) and (10-3), or graphically, using the phasor
diagram shown in Fig. 10-5. On assuming 1 0 II and
deeming VI, R, i 2 an d i 3 known in ad vance, we can find
EI , V~ and V~, and determine the per-unit voltage regulation for the secondary and tertiary windings:
D.v 2 = (V~ - VI, R)/V1, R
D.V 3 = (V~ - VI, R)/V I R
From the equivalent circuit or the phasor diagram , it is
seen that when Zl =1= 0, the referred secondary voltage de9-0169


Part One. Transformers

pends not only on the referred secondary current, but also

on the referred tertiary current (by the same token, V~ depends not only on I~ but also on I~)-a feature undesirable
from the consumer's point of view. This effect can be minimized by reducing Zl at the expense of its reactive component, X l' Practically, this
can be done by placing the
primary winding between the
secondary and tertiary windings, as shown in Fig. 10-1.
If we express the short-circuit
inductive reactances in terms
of the winding dimensions
(see Sec. 8-2):

+ al\)/3 + a12
X SC13 '"'" (a l + a 3 )/3 + a1 3
X~C23 '"'" (all + a )/3 + a 23
where a 23 = a l + a13 + a1 2'
XSCll\ '"'" (a l


Fig. 10-5, Phasor diagram of a

three-winding t.ransformer


we can see that with this

arrangement Xl becomes negative (as in the phasor diagram
of Fig. 10-5) and very small
in absolute value

(X SC12
X sc13 X~C23)/2
'"'" 2al/3
a13 a 23 = ,-a l/3 < 0
Three-winding transformers are built with their windings
differing in power ratings. A relevant Soviet standard stipulates the following ratios (as fractions of the primary



S3,R/ St,R

1 1 1

The power ratios must be the same as the ratios between

the respective referred currents. It is also required that
i, = -j~ - i; (see Fig.' 10-5). However, the sum of the
secondary and tertiary currents may exceed the primary

Ch. 10 Multiwinding Transformers. Autotransformers


and the sum of the secondar y and tertiary powers mayexceed the primary power

+ VII ; ;;;;:: VII I

8 2 + S3;;;;:: 8



The same Soviet standard requires also that for the first
power ratio
for the second,
and for the third

8 3 ~ 1 1/ 3 8 1
With any power ratio, however, a three-winding transformer must satisfy the active and reactive power balances

P1=P2+ P3+ 2J P
Q1 = Q2 + Q3 + 2J Q
where 2:.P and 2:.Q are the active and reactive power losses
in the transformer itself (see Sec. 3-8).
(i i) Split-Primar y (Spli t-S econda r y) Two-Windi n g

Transfor mers
A split primary (or split secondary) consists of two electrically isolated parts , so, in effect, such a transformer is
a three-winding t ransformer. Fro m a t hree-winding unit
proper, it differs in that energy need not be transferred
between the halves of t he split winding.
The arrangement of a transformer having one primary (1)
and a split secondary (2 and 3) is shown in Fig. 10-6. The
magnetic circuit is of the core-and-shell (five-leg core) form,
as shown in Fig. 1.5b. The halves of the split secondary
are on the low-voltage side and are wound on different
legs. The primary winding, which is on the HV side, has
two parallel paths likewise wound on different legs.
With this arrangement, magnetic coupling between the
halves of the split secondary is very loose, and transfer
of energy from network 2 to network 3 by virtue of a magnetic field is negligible. Because of this, such a transformer


Part One. Transformers

may be looked upon as a combination of two separate transformers , one coupling network 1 to network 2, and the other
coupling network 1 to network 3. If only one half of
the LV secondary, say, LV2,
is loaded, on the HV side
only one of the parallel
paths, wound on the same -leg
will likewise be loaded. Of
course, such a transformer
can transfer energy in the
reverse direction as well. Then
it will have two primaries,
LV2 and LV3, each supplied
from a separate source, and
one secondary, HV1.
Fig. 10-6J Single-phase, twoThe values of Vs and V 2
winding transformer with a split
may be the same or different.
LV winding
The values of <1>2 and <1>3 and
ofi,the refferred currents l~ and I~ depend on the relative
magnitudes of VI and the referred secondary and tertiary
voltages, V~ and V~. If




i; = i;

and eD 2 = cDs
and, as a consequence, the fluxes in the outer (unwound)
limbs of a five-leg core-type magnetic circuit are nonexistent.
In the general case , when the voltages in networks 2
and 3 are such that V2 =1= V;, the referred currents and
fluxes are likewise different



=1= I; and <D 2 =1= <D s

and the difference flux, c})2 - <1>s, has its path completed
via the outer legs. If the transformer had a two-limb core,
then, with V~ =1= 1 ; , the difference flux between the
upper and lower yokes would run outside the magnetic
circuit, and appreciable eddy currents would be produced
within the sides of the oil tank and other substantial structural parts, leading to increased eddy-current losses. This


Ch. 10 Multiwinding Transformers. Autotransformers

is the reason why it is preferable to use the core form of

magnetic circuit for split-winding transformers .
As compared with conventional two -winding transformers having one HV primary winding and one LV winding,
split-winding units offer an unfailing ad vantage in that
should a short-circuit occur across the secondary terminals,
it will draw half as heavy a current from the supply line .
This is because in a conventional two -winding transformer
windings 2 and 3 are connected in parallel and its shortcircuit impedance ZSCi23 is half the short-circuit impedance
of windings 1 and 2 (or 1 and 3) in a split-winding transformer , ZSC12 = ZSC13' Understandingly, split-winding transformers have gained marked popularity.


In an autotransformer, the primary and secondary windings

are coupled bothind uctively and conductively. In fact, it





Fig. 10-7 Autotransformer connection

has a single tapped winding which serves both primary and

secondary functions.
The auto-connection used to transfer energy from an input
networlc[at voltage V to an output network at voltage V'
> V is shown in Fig. 10-7. The figure shows two windings,
1 and 2, wound on the same core and enclosing each other
(see Fig. 1-1a). The primary is on the LV side, V, and the
secondary is connected between terminal a (X) on the
input network and terminal x on the output network in
such a way that its voltage V 2 is added to V to give V'.


Part One. Transformers

The secondary of an auto-transformer must be designed

for V or V', whichever is the higher (in the circuit of
Fig . 10-7, this is V'), rather than for V 2 , as in an ordinary
The transformation ratio ri of an autotransformer is the
ratio VIV' at no-load (1' = 0). For the circuit in Fig. 10-7,


VIV ' = El/(E l


= 1/(1


21 )

where n 2l = E 21E j = w 2/wl

Electromagnetic processes in an autotransformer can be
described, using the usual transformer equations






- Z 2 /2

2 n 2l

E, = E 21n 2 1 = -Zo/o

To them are added equations describing the circuit itself,

with the positive directions assumed as shown in Fig. 10-7:


V' = V

+. V



A phasor diagram for an autotransformer is shown in

Fig. 10-8.
For insight into" the basic energy processes in an autotransformer, we shall neglect loan d the voltage drops in
the windings which enter Eqs. (10-5) and (10-6) by assuming 1 0 = 0, I z, I = 0, and I Z2 I = O. Then,
1111 2 = V 21Vl = n 2 l
V = VI

V' = V + V 2 = VI (1 + n 21 ) = Vln
1 = 11 + 1 2 = 1 2 (1 + n 21) = 1'1n



Ch. 10 Multiwinding Tr ansformers . Autotransformers

With the above simplifications and the active and reactive

power losses neg lec ted. t the t ot al power ofj an autotrans-


Fig. 10-8 Ph asor di agr am of an autotransformer in th e case of a resistive-inductive load; cpr > 1, 1! 21 = 0.5, I! = 1/(1
0.5) = 2/3


former ma y be wr itten as the sum of two components


-+ V II

= ST(a) -+ S c
= (VI -+ V 2) 1 2

V 2I 2

VIl' = S'

-+ V II


S T (a)

= VIII = V 2 I 2

is the power transferred

power), and

inductively (the tr ansformed

Sc =

VI I 2

is t he power t ransferred from the primary to the secondary

network conductively (the condu cte d power). This is why an
autotransformer needs t o be designed to withstand only
the ST( a) term which accounts for only a fraction of t he
to tal power, S , called t he auto fr acti on
ST(a/S = T1 2I 2/V'I 2 = (V' - V)/V' = 1 - 12 (10-9)
where 12 < 1.


Part One. Transformers

In an autotransformer as in a conventional transformer,

the transformer size is sole ly determined by the transform ed
pow er S T . As has been shown in Sec. 9-1, the transformer
size is proportional t o S 1/ 4. For an ordinary transformer,
ST = S . For an autotransformer, S.;' = S T ( a) = (1 n) S.
Accordingly , given t he same power rating, an autotransformer will be smaller in size and less exp ensiv e to make.
An autotransform er becom es more attractiv e as its transformation ratio n differs progressively less from unity .
For example, when n = 0.9, the transformed power decre ases t en-fold, where as when n = 0.1 , it is about the same
as that of an ordinary transformer . This is the reason why it.
is common pra ctice t o m ake au totransformers with n ranging between 0.5 and 1. In su ch cases , the relatively mor e
expensive insulation on the secondary is more than offset
by the reduction in the weight of, and the losses in, the
Autotransformers ar e widely used to power domestic
appliances and control-system units and come in size s
from 10 to under 1 000 VA . In the Soviet Union , they are
common in hi gh-voltage power t ransmission lines where
they are used t o t ie networks operating at closely spaced
voltages, namely '110 and 220 kV, 220 and 500 kV, and 330
and 750 kV. The overall capacity of such autotransformers
runs into hundreds of megavolt-amperes.
Autotransformers may be used t o both step up and step
down the applied voltage . For example , the autotransformer
in Fig . 10-7 will step down the applied voltage, if the load
is connected to receive V , and power input comes from a network at V' .
Apart from single-phase, two-winding autotransformers
(Fig. 10-9a), power syst ems oft en employ three-phase, twowinding autotransformers (Fig. 10-9c), and also singlephase (Fig. 10-9b) and t hree-pha se (Fig. 10-9d) three-winding au to transform ers . The st andard win ding conn ections
and phase displ acement gro up s used in t he Soviet Union for
autotransformers are li sted in Fig. 10-9. Auto-connect ed
single-phase windings are deno ted by I an l a , whereas starconnected t hree-phase au to windings, with t he neutral
av ailable for conne ct ion , are den ote d by Y n. au to '
Autotransform ers may constitute an electr ic hazard ,
especially when 1/n ~ 1, because of direct connect ion
between the HV ne twork at V' and the LV network at


Ch. 10 Multiwinding Tra nsformers. Autotranslormers

v ~ V' . In t he abse nce of grounding, t he voltage between

the LV con duc tors and ground is V' /2, wh ich appea rs owing
to cap acitive coupling between the HV wires and ground.
IIV ond MV

HVond LV

~ UJ-t

I I Auto


~o /1-0-0

ta ;

HVand LV

I (6)


Yn , aulo
HVand M V



A Am B 8 m C Cm

Fi g. 10-9 Winding connec tions and phase displace ment groups for
autotran sform ers

For example, if an autotransform er were used t o ste p down

from 3 kV t o 220 V, the voltage bet ween the 220-V wires
and ground would be 3/ 2 = '1.5 kV. This is why applicable
safety codes guard aga ins t usi ng au to tr ansformers with
~/n ;> 2,

Part One. Transformers


The use of autotransformers with n ~ 1 runs into certain

difficulties because fairly heavy short-circuit currents are
likely to develop. If an autotransformer is energized with
V' on the HV side (Fig. 10-7) and a short-circuit occurs on
the LV side, V would reduce to zero, and winding 1 of the
autotransformer would be short-circuited. At the same
time, the voltage across winding 2 would rise from V 2 to
V', and this would lead to a further increase in the shortcircuit currents. On setting V = -VI = 0 in Eqs. (10-5)
. and (10-6), let us determine the steady-state short-circuit
current in winding 2:

I SC2 = V'IZsC2I= (V2IZsC21) (T1'IV2) = I sc2. T/(1-n) (10-10)

ZSC21 = short-circuit impedance with winding 1 short-circuited and with the supply voltage applied to
winding 2
I sc2. T = V21ZsC21 is the short-circuit current in winding 2
with V 2 impressed on winding 2.
It is seen from Eq. (10-10) that the short-circuit current
in winding 2 of an autotransformer is V' IV 2 = 1/(1 - n)
times the short-circuit current of a conventional transformer
used to transfer energy from a network at V 2 into a network
at VI' The closer is n to unity, the larger the short-circuit
currents, I sc2 and I SC1 = n21Isc2' and the more dangerous
are their consequences.


Transformers in Parallel


Use of Transformers in Parallel

Parallel connection of several transformers is widely used

in electrical systems. In many cases, it is the only way to
convey large blocks of power over large distances. Several
transformers operating in parallel at a major substation
cannot be replaced by a single unit of the same total power
rating, because it would be prohibitively large and unwieldy
both to manufacture and move it to its permanent location.
Even at not so large substations, the use of several transformers operating in parallel offers a more convenient way
to tackle the problems of reliability and plant expansion.


Ch , 11 Tra nsf orme rs in Parallel

Shoul d any unit fa il, the remaining ones will still be operabl e and take up the load previously carried by the faulty
transformer. In the meantime, the fa il ing transformer can
be replaced by a standby unit whose cost will undoub tedly
be sm all in comparison with that of all the installed transformers. Also, if a substation has a sufficiently large number
of t ransformers, it is always possible to combine in paralle l
as many of them as may be necessary for optimal load sharing
and energy conversion at a min imal loss (see Sec . 6-3).
The choice of a number of transformers to be operated in
parallel is both an engineering and an economic problem in
optimization . In this problem, the variables to be opti mi zed are the t otal cost of manufacture and operation of the
installed transformers . An important po int to bea r in mind
is that the cost of the energy lost and the cost of manufacture decrease] with the increase in per unit rating , wherea s
the redundancy cost increases.

Procedure for Bringing Transformers

in for Parallel Operation

To avoid li kely errors, the transformers to be operated in

parallel must be interconnected at ident ically marke d term ina ls. An example of two transformers connected for parallel


x r-


i flJ







Xcr; Sf














Fig . 11-1 Par all el operation of t wo l/i -O singl e-phase, two-winding


operat ion is shown in Fig. '1'1-1. As is seen, the ident ically

ma rked te rminals of transformers a an d ~ (A a and A 13' X a
and X (3' aa and a (3 ' :ra. and XII) ar e respecti vely connecte d t o
the same bus.


Part One. Transformers

Let us formul ate the rules for paralleling two transformers, with their load Z disconnected (that is, with switch
8 2 open). Obviously, the primary terminals of the two transformers, namely A a , A 13' X a , and X 13' may be connected
in the above way to the input network without having to
meet any additional requirements. After the primaries are
connected for VI = VIa = V 2j3, the voltages existing between the disconnected secondary terminals aaxa and aj3xj3
will be as follows:
V 2a = E 2a = V Ia/nl2a = EIa/nI2 a

V 2j3 = E 2j3 = V I j3/nI2j3 = E I j3/n l2j3
Terminals X a and xj3 may be commoned without running
any risk. However, commoning terminals aa and aj3 may
give rise to an emf across switch 8 1

E I>. = E 2a - E 2j3
Commoning terminals aa and aj3 will not give rise to any
circulating currents in the windings only if

EI>.=E 2a - E2j3=O
or when the secondary emfs are the same

E 2a = E 2j3
For this to happen, the transformers to be brought in for
parallel operation must meet the following requirements:
1. Have the same transformation ratio. If n l2a = nI2j3 and
VIa = VI 13' the secondary emfs J will be~ the same, E 2a
= E 2j3.
2. Fall in the same phase displacement (clock figure) group.
If so, N a = N 13 = N, the secondary emfs, E 2a and E 213,
will be turned through the same phase angle relative to the
identical primary emfs, E l a = E l 13 = -VI' and will
therefore be in phase


(E l a/n l2a ) exp (jON)

= (EI 13 /nl2j3 ) exp (jON) = E2j3

The above requirements are also applicable to three-phase

transformers. When they are brought in for parallel operation, connections must be made between the identically


Ch. 11 Transformers in Parallel

marked line and neutral terminals (A a and A


B a and B 13'

Ga and G 13' a a and a B' b and b B, C a and cB ' O a and 13 ) ,

If this condition is met , the secondary line emfs will be iden-

tical in both m agnitude and phase .


Circulating Currents Due to

a Difference in Transformation Ratio

Consider two V1-0 single-phase transformers ex and p. If

their transformation ra tios ar e not the same, n I 2a =1= n 12B '
and E t>. =1= 0, the circulating currents l la , lIB' 1 2a , and
1 2 13 which will appear in the windings upon closure of
switch 8 1 , can be estimated on neglecting the magnetizing
currents (Io a = 1 0 /3 ) an d writing the equations for transformers ex and p (see Sec . 3-3) for the positive directions shown
in Fig. 11-1:



+ IlaZla,
-E I B + I I BZI B'

= -E l a

1 2a =


V 2a

= E 2a

V 2B

1 2 [3Z 2 [3

- I l a n 12a ,

. . 2a

E 2 [3

1 2 [3 =

Z 2a


- I I [3 n 12 B

Also, we must consider what happens when t he two transformers are brought in parallel while the load is disconnected (switch 8 2 is open and 1 2 = 0):




V I 13

II =

I la

+ 11 [3 '



V Ia, V 2

1 2a

V 2a

+ 1 2 [3


=12 =0

Solving Eqs . (11-2) and (11-3) for the secondary circulating

current 1 2a = - 1 2 [3 gives

where Z~c[3 = Z2[3

Z I BlnT2 13 and Z~ca = Z2a
Z I a 1n T2a
are the short-circuit impedances of the two t ransfor mers,
w hen the functions of the windings are reversed .


Part One. Transformers

Using E qs. (11-4) and (11-2), we can readily establish

t?e relationship between the voltages V2 = V2a = V2/3 and
VI that exist aft er switch SI is closed

v 1 -where n 21/3

"Z 'sea +z'se

V 2

. Zse



an21/3 + Z~ e/3n 21/3

= 1/nI2/3' n21a = 1!n 12a , and for



VI = - V 21n 21
n 21/3)/2 is the mean transformation

where n 21 = (n 2la
If the difference in t ra nsformation rat io between the
two transformers is small (n12:1.lnI2/3 ~ 1),


- i 2/3

~ E!>,I(Z~e/3

+ Z~ea)

E!>, = E 2a - E 2/3 = E 1 (1!n I2a - 1InI 2/3) ~ V 2 (~n) (11-7)

is the difference between the secondary emfs given by
Eq. (11-1), ~n is the per-unit difference in t ransformat ion
ratio, and n 12 is the mean t ra nsform ation rat io.
The circulating currents defined by Eq , (11-4) and appearing at the ra ted primary voltage.] VI = V I, R , when V 2
~ V 2 , R, can conveniently be ex pressed in per-unit, t aking
the rated current of, say, transformer a as the base quantity:
l*ea ~ 1 2al1 2a,R = I l al l l a , R
~ ~nl(VBe,a
v Se/3S a, RISI3 ,R)
Vsea = 1 2a, RZ~e a1V2a , R
v se/3 = 1 213 , RZ~e I3 IV2 I3 , R

are the impeda nce voltages of t he two transformers, and

Sa ,R
S I3 . R

= 1 2a ,RV2a , R
= 1 213 , R V 213 , R

are the power ratings (rated powers) of the two transformers .

It follows from Eq . ('1'1-8) th at eve n with a small difference in the t ransformat ion ratio, the circula t ing cur rents
may be comparable in m agni tude with t he rated currents of


Ch. 11 Transformers in Parallel

the paralleled I,transformers. For example, when 'Sa , RIS f\,jR

= 1, Vsca = vscf\ = 0.05, and /),n = 0.05, the circulating
current will be
1*. c = 1 21I2 R = 0.05/(2 X 0.05) = 0.5 per-unit
To avoid hazardous circulating currents, the transformers to be paralleled may differ in their transformation
ratios by not more than 0.005.

Load Sharing Between Transformers

in Parallel

If paralleled transformers meet all the requirements, no

circulating currents will be flowing in their windings when
the load is disconnected.
Let us load the paralleled transformers by closing switch
S2 (see Fig. 11-1) and see how the load current will be



Fig. 11-2 Equivalent circuit of two transformers in parallel operation

shared. This can be done by reference to the equivalent circuit in Fig. 11-2. In fact, it is a combination of the equivalent circuits for transformers CG and ~, as given in Fig. 6-1.
The currents in the parallel branches formed by the
short-circuit impedances of the transformers

Zscf\ =

+ n;2 Z2a = Zsca exp (jqJsca)

Zlf\ + n;2 Z2f\ = Zscf\ exp (jqJscf\)

are inversely proportional to the impedances

jlalI~f\ =

Zscf\/Z sca

(ZsCf\/Z sc a)





Part One. Transforme rs

The sum of the currents gives the load current


I = I 1a
I 1fl
If the paralleled tra nsformers are fully identical and their
short-circuit impedances are the same in magnitude, Z sc a
= Zscll' and in phase angle, (Psc a = qJsc fl ' then each transformer will carry half the total current
I 1a = Il~ = 1 1/2
If the short-circuit impedances are t he same in magnitude, but differ in phas e angle , qJsca < qJscfl ' then

I 1a = II fl exp (j L1qJs c)
and the current phasor diagram looks like one shown in
Fig. 11-2. Ea ch transformer carries a current given by

I 1a = I 1fl = 11/2 cos (L1qJsc/2)

which is 1/cos (L1qJsc/2) times the current I i/2 existing when
the load is shared equa lly. Fortunately, even with the largest possible value of qJsc fl (about 90) and the least possible
value of qJsc a (about 60), their difference is about 30,
and the resultant overload does not exceed
1/cos (L1qJsc/2) :::;;; 1/cos (3012) = 1.03
Therefore, the overload due to a difference in qJsc may be
neglected, and consideration should only be given to the
difference in magnitude between t he short-circuit impedances. It follows from Eq. (11-9) that the current ma gni tudes
are inversely proportional t o the "magnitu des of the shortcircuit impedances

I 1aJI 1fl . Zscfl/Zsca

Simple manipulations give
(Iia/lill) (Vl ,RIV 1 ,R)
Zsc flI lfl,R
V1 ,R

Not ing that


I11l ,RV1, R

Ch, '12 3-Phase Transformers un der Unbalanced Load

'1 40

we can read ily find t hat per-u nit load s on t he paralleled

transformers, S 'l,a and S * ill are inversely proportional to
th e impedance voltages, Vsc a and vs cfj :
S* aIS*fl = vscfjlv sca
If vsc fj = Vsc a , the per-unit load is abo ut the same on
either of the paralle led transformers, and eac h is being
utilized to full advantage . If one carries its rated per-u ni t
load, S'I:a = 1, the ot her, too , will carry its rated per-unit
load , S*fj=1.
If, say, vsc fj > Vsc a , transformer ~ will be under-loaded,
although transformer ex is carrying its rat ed load

S* fj = (vscalvscfj) S 'I:a < 1

Conversely, if transformer ~ is carrying its rated load,
transformer ex. will be overloa ded

s.; =

(vSCfjlv sca ) S'I: fj > 1

This is the reason why the transformers to be paralleled
must have identical relative impedance voltages . (I n pr ac.t ice, t he maximum difference is allowed to be as high as
10% .)


Three-Phase Transforme rs
Under Unbalanced Load

12- 1

Causes of Load Unbalance

In the preceding sections, we discussed three-pha se t r ansform ers operated in networks with symmetrica l voltages
and ba lanced l oads. Unfo rtunately, an ideally ba lanced load
is practically nonexistent in power systems, and there is
always some degree of unbalance present . This unbal ance
increases with increasing power rating of single-phase loads
drawing their power from three-phase systems, an d is especially pronounced under abnormal conditions, suc h as tw oand sing le-phase fau lts to ground, fa ilure of one of t he phases , and the like .
To form a reliable estimate of the unbalance that m ay be
to lerated in an operating system, we need a mathematical
description of what happens in a transformer in t he case of
an un ba lanced load.

. Part One. Transformers

. In the most general case, a transformer may be not only

~ '~

carrying unbalanced secondary line currents I c .Itne- I b.\Ine

\a~nd je. llne, but also operating
from a network with unba.,
Iance di .Iine voltages 11A D , 11B C , and V CAo To obtain a
of the events taking place in such a case,
e picture
'we must determine the phase secondary currents I a' I b'
and } e (if the :secondary is delta-connected), phase and

. ..

line primary currents I A, I B, I C and I A.llne, I B.llne, I c.une

(the latter only if the primary is delta-connected), primary
phase voltages 11 A' VB' and V c (the primary is star-connect. . .
cd), :secondary phase and line voltages Va' Vb' V e and
Va b , Vbe' Vea (the latter only if the secondary is star-connected).
;), Most commonly, these quantities are found by the method
_Qf ) ymm etr ical (phase-sequence) 'com ponents. By this meth.9d;' an unsymmetrical (unbalanced) set of phase voltages,
currents 0 1' fluxes is resolved into symmetrical systems equal
in number to the number of phases and formed by the respective components in the positive, negative and zero phase
An important point to bear in mind is that the phase
sequence in the supply network has no bearing on what happens in a transformer under balanced load contitions. This
implies that its winding impedances for the negative-phasesequence (NPS) currents do not differ from those for the
..positive-phase-sequence (PPS) currents, Zl' Z2 and Zo
(see 'Chap . 8). Special treatment is only needed for zero:phase-sequence (ZPS) currents.
; 12-2

Transformation of Unbalanced Currents

. (i) Star-Connected Secondary

With the secondary star-connected, the specified unbalanced

~ Iine. currents j a, line' jb.

.\.t ime, phase currents

l lne i


Ie, l Ine

Ia=Ia.line, Ib=Ib .line,

are, at the same


Ch. 12 3-Phase Transformers und er Unbalanced Load


The phase secondary currents may be represented as sums

of symmetrical current compon ents , namely, PPS currents
i.. = (t a 2i b ai e)/3

NPS currents

ja2 = (i a
ai e)/3
I b2 = I a2a
i.; = j a 1a 2
where a = exp (j2n/3), an d ZPS currents




= i. , = jeD =
+ib i e)/3
How a set of phase currents is resolved into symmetrical
components is illustrated in Fig. 12-1.


iao!H ico

Fig. 12-1 Resolution of an unba la nced system of currents into sy mmetrical components

The neutral wire of a Yn-connected winding carries a current

. .
In = r ,
t , = 3I ao

As is seen, this current is three times t he ZPS current.

Assuming t hat t he system (network) is linear and neglecting the magnetizing currents in comparison with the load
currents, we may deal with the transformation of each of
the symmetrical systems individually.

Part One. Transformers

The relationship between t he PPS primary and secondary

currents has been established in Sec. 3-7. It has been shown
that whate ver the conn ection of the second ary and primary





~ -I~

This equation may be written for any of the three phases,

using the notation adopted for an unbalanced load:

where j ~I ' hI' and j~I ar e the secondary currents referred

to the primary side. *
Because the phase sequence in a transformer is of no importance, the relationship between the NPS secondary and
primary currents will be the same

This relation may be extended to ZPS currents in cases

where they can flow in t he primar y winding, that is, when
t he primary is star-conn ect ed, with the neutral brought
out, or del ta-conn ected:



= I BO

-i/,o = I




Thus, when the primary is connected in a Yn or /'0", its

phase .current s are equal to the corresponding secondary
phase .current s:

+ I A 2 + lAO








= -j~
When the primary is connect ed in a Y n its line currents
do not differ from its phase currents

j A,lIne =

iA ,

j B,lIne =


i C ,lIne


* Equation (12-4) and all th e other eq ua tions in this section ar e

written for th e winding connec tio ns where the id en ti cally marked
phase wind ings (A and a, Band b, C and c) are wound on the sam e-leg.


Ch. 12.3-Phase Transformers under Unbalanced Load

andthe current in the neutral wire

jN = si AO= - sjao -


is equal to the referred current in the neutral wire on t he

secondary side ,
In a delta-connected pr imary, t he li ne currents do not
contain ZPS components


- (I Bl + I BZ

a.. + jAZ) -


+ h2)


+ j B2)
+ (hi + h 2)


The ZPS current l AOhas its path completely around the

delta and does not appear in the line wires (Fig. 12-2).





; -Ib



B (i:-i~)


(jk-i~ o)


u-u." )

A 'tu


c I' =3Il1.o

i n=ajl1.o






-t;ir :













Fig. 12-2 Transformation of unbalanced currents by various winding


In a star-connected primary, there is no neutral wire

that might carry ZPS currents. Therefore, no ZPS currents
into the primary
. . induced


Part One. Transformers

and the ph ases of this win ding onl y carry PPS and NPS

t, =



i., + i A 2

= -j~l -

h2 =

-(i~ - i~o)

i, B) . j Bl + j B 2 = -ib1 - Ib2
-ii; = -(j;, - h o)


As follows from t he foregoing, wh ate ver the form of conn ect ion , the PPS and NPS curren ts are transformed identically . Therefor e, it appears reas onable t o treat separately only
the ZPS phas e curren t s, and t o lump t ogether the PPS and
NPS currents


I a = I( a) + l ao

where I(a ) = I al
I a 2 is the PPS and NPS currents in
phase a, shown by the da shed line in Fig. 12-1.

(ii) Delta-Connected Secondary .

When t he secondary is connected in a delta, t he specified
line currents always sum t o zero

I a ,line
I b,lIne
I c,line = 0
Because the line currents are t he differences of phase currents




la, line =Ia-I b, I b,lIn e =Ib-I c, I C ,line=Ic - Ia

and the phase cur re nt s do not contain ZPS components and
sum to zero
t ; = 3I ao = 0
we may wri te the phase currents in terms of t he line currents

i a = ( t,llne - j c,llne)/3
i, = (i, , llne - i, .llne)/3
i. = (j c,lIne - i blI ne)/3



Gh. 12 3-Phasc Transform ers under Unbala nced Load

Graphically, the phase currents are defined by the centroid

of a line-current triangle, lying at t he intersection of its
medians. As will be recalled from school ma thematics, the

-r .

a ~----',------=;;P}c




Ie,line I


rIa, line



" ":


, -'

\ 1

Fjg, 12-3 Determining th e ph ase currents in a delta- connected windnig


, . j ").

intersection of the medians in a triangle li es two -thirds ..qf

the way from its apexes (see Fig. 12-3).
Because the secondary phase currents do not containany
ZPS currents , they are fully transformed into theprim.ary,
no matter how it is connected (see Fig. 12-2):
. . s>


j B = - j b' i C = -j~
When the primary is star-connected with its neu trrl;!l
brought out,

jA =

- : " :~ I

t N = 3j AO = -3ho = 0

.i .

". : , . ' 01


i :: '~.

", ' 'HC'<:

Magnetic Fluxes and EM Fs under
Unbalan ced Load Conditions

Under unbalanced load conditions, the total magnetic

flux may, in a linear approximation, be visua lized as ~the
superposition of the fluxes set up by PPS , NPS :" and ; ZPS
currents (Fig. 12-4).

: <: } " J J

The balance between the PPS primary cui~erit~:aJ~ ~,

I Bl' I CI) and the PPS secondary currents (j aI',
is never complete. The unbalanced fractions of the .BPS
primary currents, which are the magnetizing curr~nts:'1B

Al;J 'cD


..~ .~~ 1

Part One. Transformer:'!


hI' JBI hI' JCI i.. give rise to a balanced set of

PPS fluxes <hAl' <h El' and <b CI (see Fig.H2-4). This system of fluxes has all the properties of flux~s in three-phase


Fig. 12-4 Balanced components of magnetic fluxes and emfs und er an

unbalanced load

transformers under a balanced load (see Sec. 4-1). What is

especially important is that these fluxes sum to zero.

cD A l

+ (DEl + cD

c I

= 0

So, they are free to traverse a closed path in any form of

magnetic circuit.
The same goes for the systems of NPS currents in the
primary winding (1A2 ' 1 B2' 1 C2) and in the secondary winding (I a2' 1 b2, 1 e2)' They, too, are not completely balanced
and form a symmetrical set of NPS fluxes (see Fig. 12-4)

cP A

+ cP + cP


In contrast, the ZPS fluxes established by the ZPS currents and their paths substantiallydependjon how the windings are connected and the form of the magnetic circuit.
As has already been explained in Sec. 7-3, the ZPS fluxes
have their paths completed within the magnetic circuit
only in the core-and-shell (five-leg core) type of transformer
and also in a three-phase bank of single-phase transformers. In a core-type transformer (see Fig. 12-4), the in-phase


Ch, -12 3-Phase Transformers under Un ba lanced Load

ZPS flu xes

<D A O =

cD B o

= <D c o = <Do

have their paths outside the magnetic circuit and within

the nonm agnet ic ga ps, tank sides, an d ferro magnetic structural par t s.
Because the ga ps offer a high opp osi tion, the ZPS flux es
in a core-type transformer are m arkedly sm alle r than t hey
are in a five -leg core-type transformer or in a bank of transformers (With the same mmf, I a ow 2 ) .
The ZP S fluxes are especiall y strong when the ZPS currents flowing in a star-connected secondary with its neutral
brought out are not balanced by the currents in the primary,
which usua lly happens when the latter is star-connected
with no neutral wire available (see Fig. 12-4).
As with PPS and NPS currents, it appears reasonabl e t o
lu mp together the PPS and NPS flu xes and t o treat sepa rate ly only t he ZPS fluxes :


<D (C)


cD c I

CV( C)

+ cD"

cD( B ) = CV BI + cD B2 ' and

<D C2 are the sums of the P PS and NPS

= .cD Al

CV c =

(1) A2 '

Sinusoidal PPS and NPS fluxes induce in the primary
phases the mutual emfs of positive and negative phase sequence s [see Eq . (3-7)]:
E( A ) =


E (B) = E B I

+E = -

(wll!2) w lcD(A) =


(wll! 2) wj(1) (D) = E(b)


B2 =




where Eia) = E(a)wI /w2 is the PPS and NP S mutual emf of

phase a referred to the pr imary sid e.
Sinusoidal ZPS fluxes induce in the primary phases mutual emfs of zero phase sequence (see Fig . 12-4):


(w/ 11 2) wjcDo= E~o


where E' aO = Ea ow I /w 2 is the ZP S mutual secondary emf

referre d to the pr imary side .
The ZPS mutua l emf may be expressed in t erms of the
ZPS currents
set ting u p the ZPS flu x cD o (the ZPS flu x


Part One. Transformers


needs to be treated separately only when the primary is

star-connected and it carries no ZPS current, as in Fig. 12-4)


Roo jX oo is the mutual impedance to ZPS currents

X oo = Ulw;Aoo is the ZPS mutual reactance proportional to
the permeauce for the ZPS flux
Roo = resistive component of the mutual impedance, due
the hysteresis and eddy-current losses in the ferromagnetic structural parts, associated with the sinusoidal ZPS fluxes.
Because t aO is the magnetizing current for the ZPS fluxes,
Eq. (12-16) is written by analogy with Eq. (3-7) defining
the relation between 1 0 and E 1


Dissymmetry of the Primary Phase

Voltages under Unbalanced Load

The equations defining the primary phase voltages under

unbalanced load are written by analogy with those for the
balanced load conditions, Eqs. (3-13). The mutual emf E 1
is replaced by the mutual phase emf which is the sum of
the PPS and NPS emfs, and the ZPS emf, E AO = E BO
= E c o:

VA = -E(A) E AO


Vc =

-E(B) E A o + ZII B
-E(c) - E A O + ZlI c


The primary line voltages V AB, V BC and V CA, which are

in the general case unbalanced, are specified in advance.
When the primary is delta-connected, the primary phase
voltages are the same as the specified line voltages and
need not be determined. Also, the ZPS current lAo around
the delta balances the secondary ZPS currents, and EAO
in Eq. (12-17) vanishes.
When the line voltages are symmetrical, the phase voltages
in a delta-connected primary are likewise symmetrical.


Ch. 12 3-Ph ase Tra nsf orme rs under Unbalan ced Load

When the primary is s tar-connected with its neutral wire isolated, the specified line voltages are the differences of the respective primary phase voltages



Also, adding tog ether the right- and left-hand sid es of

Eqs . (12-17) and recalling that t he emfs an d currents containing no ZPS currents sum to zero

+ E BI + Ecl )
+ (E A2 + EB2 + EC2) = 0
I iA) + I(B) + I (C) = (I AI + I BI + I cl)
+ (I A2 + I B2 + I C2) = 0

E( A)

+ E (B) + E (c ) =


we obtain an important equat ion

VA + V B + VC

= - 3E AD = 3h oZoo


Sub tracting t he secon d line in Eqs. (12-18) from t he firs t

and re calling Eq . (12-19), we get



+ V B + V c) + 3V B
3 (E AD + VB)

V BC = -(VA

VB =

CV A B- VBc )/3 - EAo =

V (B) -


and by analogy,

= CrT BC -

VcA )/3 - EAD =


V ec ) -


et c. (12-20)

Here, V( B ) and V( c) are t he phase voltages with no ZPS current flowing in the secondary, t hat is , when I c o = 0 and

E AO = -Zoo/Ao

As. is seen from Fig. 12-5, the phase voltages V( A), V(B),
and V(C) are directed away from t he centroid of the line':'
voltage triangle, N, towards its apexes [see also Eq. (12-13)
and Fig. 12-3 for phase currents].


Part One. Transformers

_ When the line voltages form a symmetrical set, V AB

.VB C = VCA' and there is no ZPS current flowing in
the secondary, i ; = 0, the primary phase voltages are
likewise symmetrical

V( B)

V( c)




The appearance of ZPS currents ii a O =1= 0) causes the centroid of the line-voltage triangle to shift by a distance Eo
(from point N to point N 0)
and upsets the balance of the
, .J11
voltages. Now, even if
the primary line voltages are
balanced the phase voltages
will be unbalanced, V.4. =1=
VB=I=V C '
In core-type transformers,
the primary phase voltages
c are distorted considerablv
t: &t.~-:~~~---less, because the reluctanc~
to the ZPS fluxes is many
times that existing in a fiveFig. 12-5 Phasor diagram for
primary voltages under an unleg core (shell-and-core) type
balanced load
or in banks of single-phase
It follows from Eq. (12-20) that the unbalance of the
phase voltages may arise from the dissymmetry of the line
voltages even though there are no ZPS currents flowing. As
regards the symmetry of phase voltages, it is preferable to
connect the primary in a delta, because, given symmetrical
line voltages, the phase voltages will not be distorted even
when the secondary carries a ZPS current.


Dissymmetry of the Secondary

Voltages under Unbalanced Load

The equations defining the secondary voltages may be written by analogy with Eq . (3-19) applicable to balanced load.
The referred mutual emf E~ = E 1 is replaced by E(a) = E C

Ch. 12 3-phase Transformers under Unbalanced Load

or E( Bl, and E (Cl, and also

-11~ =

-E( A l -

E~ o =

EA O +


E A O:


-v;, = -E(B) - EA O + ZJ;"



Eliminating the emfs between the above equations by

invoking Eq. ('12-17), we can express the secondary phase
voltages directly in terms of the primary phase voltages:



- 11;'

VB -

ZJ + z~h,


It may be added that Eqs. (12-22) are applicable to any

form of primary and secondary winding connection.
If the primary is delta-connected and the secondary is
star-connected with its neutral brought out, the ZPS se

condary current I aD is balanced, from a magnetic point of

view, by the primary ZPS current lAo flowing around the
delta, there is no ZPS flux, and the primary currents are
equal to the respective secon dary currents referred to the
primary side:


= jAl + j A2 + jA o = -j~l -

= -i;,

= -j~

hz - ho

= -I~

Now the referred secondary phase voltage differs from the

primary phase voltage by a relatively small voltage drop
across the short-circuit impedance (as under balanced-load

-11;' =

VB + zsci;,


where Z sc = Zl
Z~ is the short-circuit impedance .
When the specified primary line voltages are symmetrical ,
the primary phase voltages are, as already noted, likewise
symmetrical, and the dissymmetry in the secondary phase
voltages due to a dissymmetry in the currents is relatively

Part One. Transformers


Equations (12-23) may be used to determine the secondary voltages also when the primary is star-connected, and
the secondary is delta-connected (a Y/!1 transformer), because then the secondary and primary currents contain no
ZPS currents (I AD = I en = 0). In the circumstances, the
primary and secondary currents balance themselves as well

-hI -h2 =

i A =jAI +jA2 =


I B = -Ii,


whereas the ZPS flux and the ZPS emf reduce to zero. Therefore, [see Eq . (12-20)1 the primary phase voltages are
determined by the position of the centroid of the linevoltage triangle





c =


and, given symmetrical line voltages, are themselves symmetrical, and Eq. (12-22) reduces to Eq. (12-23).
. The ZPS currents I aD may cause a more noticeable dissymmetry in the primary and secondary phase voltages when
the primary is star-connected and the secondary is starconnected with its neutral brought out, and there is no
ZPS current flowing in the primary [see Eq. (12-10)1:


= i(A) = i AI

i; = hI + j~2

+ i A2 =

+ j~o

-hI -

= j,a)

= -j,a)

+ j~o

In view of Eqs. (12-10) and {12-20), we may re-write

Eqs. (12-22) as
- V~ = (V(A)

+ h1Zsc) + j~2ZSC + j~oz~

- Vi, = (V( + h1Zsc) + ib2Zsc + hoz~, etc.
where Z~ = Z~o + Z~ is the ZPS impedance of the



secondary referred to the primary side.

When the primary line voltages are symmetrical, V(A)'
V( B) and V( C) are likewise symmetrical, and the dissymmetry is related to the voltage drops due to the ZPS currents (j a2Zsc, ji,2Zsc, j~2ZSC) and the ZPS currents, I~oZ~.

Ch. 12 3-Phase T ran sformers und er Unb alanc ed Load


As is seen from Fig. 12-6, the voltage drops due to the

PPS currents (j~ tZsc , hnzsc, j~lZSC) do not lead t o a voltage
It is to be noted, however , that even in a core-type transformer in which t he ZPS impedan ce is rela tivel y small
(Z;~n = 0.3 to 1.0) and onl y sev eral times the per-unit shortcircuit impedance (Z*sc = 0.05 to 0.13), the unbalance o~

Fig. 12-6 ~Phasor diagram of a YIY 11 transformer und er an unbalanced

load (the primary line volt ages are balanced)

phase voltages is more noticeable due to ZPS currents than

to NPS currents of the same magnitude. The effect of ZPS
currents is especially troublesome in shell-and-core (five-leg
core) transformers and in banks of Y/ Y11 tra nsformers. This
is because the ZPS flux es h ave their paths wi thin the
magnetic circuit in the same manner as the PPS fluxes.
In such transformers, Z~n = Z,j;O= -10 t o 100, so even small
ZPS currents gi ve rise t o a prohibitive dissymmetry of
phase voltages. This is why the Y/ Y 11 connection ought not
to be used in shell-and-core (five-leg core) transformers and
in banks of t ransformers. In core- type transformers using
the Y/ Y 11 connection, it is important to limit t he ZPS currents.
Subtracting the second line in Eqs. (12-24) from the first,
we will find t hat t he diss ymmetry in the second ary line

~ .


Part One. Transformers

voltages is solely related to the NPS currents

V ab

Vb 11(A ) -



ir( + (hI

hD Zsc + (j~2 - i b2) Zsr ,

The diss ymmetry of phase voltages due to ZP S currents in

shell-and-core transformers, banks of transformers, and
large core-type transformers using the Y/ Yn connection can
be minimized by providing an additional com pensating
delta-connected winding. The ZPS currents ind uced in the
delta will damp out ZP S fluxes, thereby substantia lly reducing t he unbalance of phase voltages in the main windings. Somet imes, t his delta-connected winding is use d as
a tertiary winding and connected to the network. In suc h a
case, the delta-connected winding is designe d not only t o
ba lance out ZPS currents, but also to transform some power
into the network to whi ch it is connecte d.

* 12-6

M easurement of the ZPS

Secondary Imped ance

The secondary im pedance to ZPS currents is measured by

producing ZPS currents , l ao = I , in t he secondary phases.
The simplest way to do this is by series-connecting the
secondary windings into an pen delta (Fig . 12-7a).

(a, )

i ao' 1.a,


1.1 Wa


l ao
.~ oo



Fig. 12-7 Measurement of ZPS Impedance:

(a) test set-up; (b) equ iva lent circuit for ZPS current

Once the voltage, current and active power are measured

by t he instruments connected as shown in the diagram, we
can readil y find the phase im pedance to ZPS currents
Z n = F/31,

R; = P/312.,


V Z~


Ch. 12 3-Phase Transform ers und er Unbalanced Load


The rea cti ve component, X n , is t he sum of the ZPS mutual

inductive reactance, X 00' and the leakage inductive reactance, X 2' of t he secon dary :


= X oo + X 2 =

where A oo and A (J 2 are th e

flu x and th e leakage flux ,
The resisti ve com ponen t
R n is the sum of the ZPS
secondary resistan ce R 2 :



+ WA (J2W ~

perm ean ces to the ZPS mutual

resp ec ti vel y.
of th e ZPS mutu al imped ance,
mutu al resistance Roo and the



If, in addition t o Y/ Yn-connect ed windin gs , a transform-

er has one more winding delta-conn ected, then , relative

to the ZPS emf , it may be cons idere d short-circuite d. The
ZPS current appearing in the della, I fl , m arkedly reduces
the ZPS fluxes an d impedan ce (see Fig. 12-7a). In such
case, the ZPS impedance should be measured , with switch S
closed (see Fig. 12-7b).
When the delta is open, the ZPS impedance of the secondary winding (see th e equi valent cir cu it ) is

+ Zoo

Zn = Z2

When the delta is closed, it is su bstantiall y reduced to

Z n6.

= Z2+

Z6 Z 00

Z 6. + Zoo

~ Z2

+ Z~ Zn

because Z~ Zoo.

* 12-7

Single- and Two-Phase

Unbalanced Loads

Single-phase load, Y/Y o or Y/MY n connection (Fig. 12-8a).

The quantities specified in advance are the primary line
voltages V A B = V Be = V CA = V1 ,ll ne, and the load
impedance Z.
Phase a carries a curren t I a which has its path through
the load impedanc e Z. The other phases carry no currents:
The ZPS current

j aO
11-01 6 9

to be


(ja + j

+ j c)/3 =

j a/3

Part One. Transformers


The .sum of the PPS and .NP S currents in .phase a is


i: + i: = i; -

j(~) '




The currents in the primary winding [see Eq. (12-10)] are

" ;.



--:-i(b) = i;, -

= -j~/3 = j c

The voltage across the load ' impedance Z and the 'ph~se 'a


... . .






: wz

. c ,



_ Va



d ' . . ..r



t::t::: . ',
. (6)

, i


i a Va

ia,line ' ,


Fig. 12-8 Single- and two-phase loads in various types of connection

, 'f .:

[see Eq. (12-24)] is


iT A + (hi

+ j~2) Zsc+ hoz~ =-z'i~

where VA = Vll n e /3.

.i. Recalling' the relations' between currents, ' the 'load current
'is ' found to be "
. :, "
. .'

I~ =
: ~ r.

.. ..


~ 3V AI (2Z sc

+ Z~ + 3Z')

-' -



' 'I~ = -V3'Vll~e/ I:2Z sc + Z~ +3i, ,.


On setting Z' = 0 in Eq. (12-25); we obtain an equation

for a single-phase short-cirouit cllrrent. . '. '.' . . ' '.. '
Single-phase load, MY n eonneetlon] (Fig. 12-8b). The
quantities specified in .advance. ara .the primary phase vol-


Ch. 12 3-phase Transformers under Unbalanced Load

tages which are the same as the phase voltages, V A = V CA

= VB = V A B = V C = V BO = V li ne , and the load impedance, I Z I.
The primary currents are given by Eq. (12-10) :

t, = -i~, i B =

= 0

The load voltage is equal to the phase a voltage [see Eq,


-V~ =

VA + zscj~


The load current is found to be

i d = VA/(ZSC + Z')



VI,line/ I Zsc

+ Z'


On setting Z' = 0 in Eq. (12-26), we obtain an equation for

a single-phase short-circuit current (which, in our example,
does not differ from the current flowing in the case of a
balanced three-phase short-circuit).
Two-phase load, Y/Y connection (Fig. 12-8c). The =:quantities specified in advance are the line voltages V A B
= V BC = V OA = Vll ne , and the
. load impedance Z. Phases
a and b carry a load current I b = - I a' whereas phase c
carries no current, Ie = O. The primary phase currents definj B = -i;"
ed in Eqs. (12-10) are t, = -j~ =
1 0 = O.
The load voltage is equal to the line voltage V a b [see
Eq. (12-23)1:


V~b =

V;' -

V~ =

V + Zsc (i~ - j;,) = zi;


The load current is .given by


CV A- VB)/(2Zsc + Z')



-VA B/(2Zsc + Z')

VI, lIne/ I zz.,

+ z'


On putting Z' = 0 in Eq. (12-27), we obtain an equation

for a two-phase short-circuit current.
Two-phase load, MY connect ion. If the primary is deltaconnected and the load is arranged as shown in Fig. 12-8c,.



Part One. Transformers

the load current may be found from Eq. ('12-27).

Noting that in a delta-connected winding 11A

= VI,line and I VA - VB I =
of load current is found to be

V 3VI,line,




Ib = V 311 1,lInel I zz., -I- z- I


On putting Z' = 0 in Eq. (12-27), we obtain an equation

for a two-phase short-circuit current.
Two-phase load, Y It-" connection (Fig. 12-8d). The line
load current is I a, line = I c.Itne i whereas the line current in
phase b is zero. According to Eq. (12-13), the secondary
phase currents are

i a = (t,lIne - i c,lIne)/3
i b = -ia,Hne/3
i, = -t.s;


The load voltage is equal to the phase a voltage, Eq,

(12-23), or the line voltage V ca:

-V~ = -V~a =




= -i a ,lI neZ '

The load current is given by


i a,lIne

= -

VAI(2Z sc/3 -+ Z')

I a.lIne = V 1,lInei (V "3 I 2Z sc/3 -I- Z' I)


On setting Z' = 0 in Eq. (12-29), we obtain an equation

for a two-phase short-circuit current.


Transients in Transformers


Transients at Switch-On

Each time a change occurs in the load or the primary voltage, a transformer does not reach a new steady state until
all transients die out. Sometimes, the currents accompanying
the transients may exceed their steady-state values manyfold. The winding temperature and the emfs, all of which
are current-dependent, rise substantially and may even


Ch. 13 Transients in Transformers

exceed the maximum safe values. Obviously, if the designer

fails to take a proper account of the transients that are likely to occur in a transformer, he will not be able to choose
the correct dimensions, service conditions, and the extent
of protection needed.
To begin with, let us consider the transients that occur
when a transformer is just switched on.
Suppose that the secondary winding is open-circuited
(i 2 = 0). At time t = 0, the primary is switched into a
supply network (or system) with a phase voltage

V 1, m cos (rot

+ 1p)

The transients in the primary circuit of a transformer can

be described by a nonlinear differential voltage equation
i oR 1 + W 1 d<l>/dt = V1
where i o is the transient no-load current , <l> = f (io) is the
mutual flux which is a nonlinear function of i o (see Fig. 2-3).
Since i oR 1 ~ W1 d<l>/dt, we may, without committing a
serious error, write i o in terms of <l> as
i o = w1<l>/L O
where L o = const is the mean primary inductance
d<l>/dt = (f)R1/L o = V 1/W1


The solution of a linear differential equation with constant coefficients is the sum of two terms, a free (or transient)
term and a forced (or steady-state) term. In our case,
(f) =

o, + <l>ss

The transient term

<l>t = C exp


is the general solution of a homogeneous equation

d(f)/dt + <l>R 1/L o = 0
where -R 1/L o = ~o is a root of the characteristic equation.
The steady-state term
<l>ss = <l>m sin (wt + 11)), where <l>m = 111 m / ro w ]
is the mutual flux <l> [see Eq. (2-9)1 which is estahlished in
the transformer core at no-load and ['1 = 111m cos (wt + 1p).
The constant C is determined from initial conditions, ,_~


Part One. Transformers

If we neglect the residual flux, eD res = 0, then at t = 0

the flux in the core is zero:

cp 1=0 = eDt

+ eD ss

= C

+ eD m sin 'ljJ =



- c:D m sin 'ljJ exp (-aot)

+ c:D

sin (rot

+ 1p)


The worst case at switch-on occurs when 'ljJ = +n/2,

because at t = 0, VI = O. In the circumstances, the initial



<P= <1'y





i omar


f 2.3 lj





Fig . 13-1 Variations in the flux and magnetizing current of a transformer at switch-on; 'IjJ = -n/2, cD core = cD exp (-aot)

value of the t ra nsient flux is equal t o t he peak value of

the steady-state flu x , C = +eDm , and, as is seen from
Fig . 13-1, a half-cycle after switch-on the flux in the core
rises to a maximum va lue equal to twice the peak value,
eD ma x ~ 2c:D m . (By this instant, the transient term subsides
very little, exp (-aon/ oi) ~ 1.)
The current in the winding following switch-on can be
found graphically from Fig. 13-1, using the magnetization
curve in Fig . 2-3. *
The maximum switch-on current io, max observed a halfcycle after the onset of -t he transients may exceed the peak

* The replacement of the nonlinear magne tization curve by a linear one results in an enol' in the va lue of a o and the damping of the
transient flux .

Ch. 13 Transients ' in Transformers

value of . the -rat ed load current

i o, max~ l!2 It.R~ l! 2 I o;w

This point mU1'1t be borne in .mind when adjusting the setting

of the protective relay's and carrying out' an open-circuit
.'. .
. ..
:. .


Transients on a Short-Circuit' Across

the Secondary Terminals

,~ l

~ .

We shall 'li~it ou~~elves to a baian~ed (three-phase) shortcircuit across the secondary terminals of a transformer.
Suppose that prior to a short-circuit, that is , at t < 0, .the
primary was energized with
. V1,m


cos (wt

+ ')


and the secondary was open-circuited.

If we ignore the magnetizing current and deem i l = -i~,
the transients on a short-circuit may be solved, using an
equivalent circuit for a short-circuited transformer.
Referring to Fig. 5-2b, the equivalent circuit of a shortcircuited transformer contains a resistive ' component,

R se = R I

+ R~

and an inductive component

+ X~
= ' 1' + ;L' ~

X se = X 1

such that

X'scl (0

~ La c

: :' Thetrarisfents in s~ch 'a circ~it . can - be 'desoribed 'by " ~

linear differential equation with constant coefficients; R se
- ' 'const ant , and .Lse . constant: '
" :'

Rsei l

+ Ls~ dit/dt =


'whose solution is
The transient term
it,l = C exp


is the solution of a homogeneous ' equation


+ Rseil/Lee

= 0

. (13-4)

Part One. Transformers


where - R sel L se = ase is a root of the characteristic equation.

The steady-state term
ii, ss =

1/ 2 [se cos (cut +

fp - fPse)

is the particular solution of Eq. (13-4) at t = 00, or the

steady-state short-circuit current. The amplitude and phase
of this current can be ascertained from a short-circuit
equivalent circuit (see Fig. 5-2b):

V2 [se =

V imlV R~e + (cuLs e),l = Vim! I Zse I

fP se = arctan (cuLseIR se)

The constant C may be defined from initial conditions.

Prior to the short-circuit, the transformer was running at
no-load, so (neglecting the no-load current) we may deem
that at t = 0,


= -


= C+



= ii, t + ii, ss

V2 [se cos (11; - fPse) =

V2 [se cos N' -


and the transient current is

it = - i~ = - l!2 [se cos (11; - fPs e) exp (-as et)

+ l!2[secOS(cut+fP-fPse)


The transient time is in fact the time required for the

transient current iI,t to die out. In timet = 1!a se after the
onset of the transients, the transient current falls to lie
of its original value. In time 3la se, it falls to lle 3 , or onetwentieth of its original value and is practically non-existent.
The time required for the transient component to reduce
to 1!e of its original value is called the transient time or
decay modulus, r = l la.
For power transformers (see Sec. 3-6),
'r se = Vase = Xse/cuR se = 0.01 to 0.2 s
that is, the decay modulus increases as the power rating
of transformers goes up.
An increase in the initial value of the free (transient)
com ponent leads to an increase in the short-circuit current,


Ch, 13 Transients in Transformers

A short-circuit is most severe if it occur s when til = (Psc or

CPsc + rr. In such cases, the initial value of the free
component is equa l to the peak value of the forced compo nent
- 11 2 I sc cos (1/-1 - (Ps c) = + V"2 Is

tp =

The waveforms of short-circuit currents when til = (Psc + n

are shown in Fig. 13-2. Th e currents in the windings attain


Fig . 13-2 Var ia ti ons in transform er cur re nt du rin g transients followa sh or t-circuit on the secondary si de (lp = qJsc

i n~

their maximum values a half-cycle after the onset of the

ij, max = I i~ Imax =

112 I sc [1 +exp ( -

n /coLsc)!

Dividing t his current by the peak rated current and assuming that the primary voltage is at its rated value, we get



11 2 i ; R = (Is c/IR) [1 + exp (-


= (tlv sc ) [1 + exp (- n/coLsc)]


where vsc is the per-unit short-cir cui t current.

In power transform ers, the maximum short-circuit current
ma y be as high as
ij, maxi V 2IR = 25 t o 15 per un it


(The larger va lues app ly to transform ers of lower :power

ratings .)


Part One. Transformers

A transformer must be designed so that a short-circuit

would not put it out of service or cut down its service life.
In choosing the winding design and clamping arrangement, preference should be given to those preventing"the
damage that the electromagnetic forces might do to the
windings during a short-circuit. The point is that the .prlmary and secondary windings of a transformer carry currents
flowing in opposite directions. On a short-circuit, i l . max
and i 2 max are many times the rated values. The electromagnetic forces arising from the interaction of the oppositely
directed currents in the primary and secondary tend to
compress the turns in the inner winding and to expand the
turns in the outer winding. The electromagnetic forces
arising from the interaction of currents flowing in the same
direction in the turns of each of the windings tend to compress
the windings in height.
During a short-circuit, the electromagnetic forces, which
are proportional to the current product, t-: maxi2. max' or
the square of each current, i; .max or
max, increase 225 to
625 times. These forces are pulsating at a frequency 2/
= 2 X 50 = 100 Hz without reversing their direction.
The windings will not be damaged if the accompanying
mechanical stresses do not exceed 50 to 60 N mm -2.
No less dangerous is the heat effect of short-circuit currents, because the copper losses (proportional to the current
squared) increase many-fold, and the temperature of the
windings abruptly rises. Since the free component of the
short-circuit current decays in 3..sc = 0.03 to 0.6 s, the
rate of temperature rise may be evaluated from the steadystate short-circuit current. This current causes the copper
loss to grow 49 to 225 times.
The current density in the windings builds up appreciably
and may be as high as 20 to 40 A mm -2. If the windings are
assumed to be heated adiabatically (that is, the heat liberated within the windings is not transferred to the surroundings), the temperature of the windings will rise at the rate
given by
J2/170 ~ 2.4 to 9.5 C S-1


Prior to a short-circuit, the maximum safe temperature

of the windings may be as high as 105C (see Sec. 16-1). The
short-time maximum safe temperature of the windings, at
which their insulation still remains intact, is set at 250C.


Ch. 14 Overvoltage Transients in Transformers

If we know the rate of temperature rise, we can readily

find the time, t sc' during which the winding temperature
will go up from 105C to 250C:
t sc

2.5 (100vscIJR)2 = 5 to 25 s

As a rule, the protective relay(s) will disconnect a transformer from its supply much earlier, and the winding temperature will not rise to it s limit.


Overvoltage Transients
in Transformers


Causes of Overvoltages

In service, transformers are often subjected to overvoltages. For example, an overvoltage may deve lop when an
element (or elements) of an electric system is turned on or




//} '/./

Fig. 14-1 Overvoltage waves

off (switching voltage surges), but this contingency is

usually provided for in t ransformer design.
More dangerous to transformer insulation are overvoltages produced by lightenings which strike line conductors
and induce high-voltage waves in them (lightening voltage


Part One . Transformers

An overvo ltage wave propagates both ways from the point

of occur rence at a velocity very close to that of light. An
overvoltage wave has the shape of an overdamped pulse
with a steep leading edge (Fig. 14-1a) . The rise time of this
wave is usually a split microsecond or millisecond; the pu lse
duration equal to the wave length runs into tens of microseconds . Figure '14-1a shows a standard total overvoltage
wave used in testing transformers for pulse strength. Its
length, in terms of the fall or decay time to half its peak
va lue, is 50 X 10- 6 s.
To minimize overvoltages (see Fig. 14-1b) , it is usual to
equip transformers with a spark-gap , SP, which will break
down at V m - Ahead of a spark-gap, the overvoltage wave
may have a very la rge peak value , V mo- Past the spark-gap,
its peak value, V m, ought not to exceed the voltage used in
testing the insulation for pulse strength (see T able 14-1)*.
Ta ble 14-1 P ul se Test Voltages for Tranaior mer
Insulation used in the USSR
Windi ng voltage
class, kV


Peak value of test

voltage (total wave),
44 60 80


35 110 115 220 330 500

108 200 480 550 750 1050 1550

The transients that take place in a transformer when the

incident wave has a fast rise time, t; (a few microseconds)
are of a very complex character .

The Differential Equation for

the Initial Voltage Distribution
in the Transformer Winding

The transients occurring in a transformer as the ap plied

voltage is raised from zero to V m during the rise time t;
may be likened to what ha ppens when the same transformer
is energized with an alternating voltage having the same
peak va lue, V m' an d a peri od T t = 4t r (see t he dashed line

* The wave reflected f rom th e input ca pacitance of the transformer is combined with the incident wave, hut tho total wavo does not
raise the voltage above V m at which the spark-gap breaks down .


eh. 14 Overvoltage Tr ansients in Transformers

in Fig. 14-1a). This alternating voltage has a fairly high

frequency . For the sta ndard test wav e, this frequency is
f t = (jJ t/ 2n = 'li T t = 'lI4tr
= 1/(4 X 1.2 X 10-G)
= 2.08 X 105 Hz
At such a frequency, we may no longer igno re the capacitive
coupling existin g between t he wind ing elements, Cd, and
betwe en the winding elements and the grounded parts of the

. . . ----,r, -

- ....









. X



..L C'
,0sit ,0sit














.J... e'"
Sh ,



- L'f.k





F ig . 14-2 Overvoltage in a windi ng: (a) cross-section al view; (b)

equivalent circuit :
1-winding conduct ors ; 2-grounded parts; 3-through 6-electrostatic shields around the coils next to th e winding start; Cd-capacitance
between coils; Cq-capacitance of coil to ground; C~h , C ~;J, C~ l; - cap acitances between shi elds and winding

transformer, C~ (Fig. 14-2a). Nor may we use the usual

equivalent circuit which only considers inductive coupling
and is applicable to operation at the rated frequency (which
is 50 Hz for power t ransformers in t he Soviet Union).
Now we must use an equivalent circuit which includes
both the inductances of the various winding elements, L ',
and the capacitances between t hem , Cd, and their capacitances to the grounded parts, C~. Such an equivalent circuit, wi th the te rminal X grounded. is shown in Fig. 14-2b.


Part One. Transformers

If we are only interested in the initial voltage distribution,

when the voltage at the winding start is V m' then, because f t
is very high, we may deem that no currents can flow in the
turns because their inductive reactance co tL' is very high.
Rather, they flow in the series capacitive reactances, 1!roCd,
and the shunt capacitive reactances, 1/roC~. Therefore, we
put ro tL' equal to infinity
and assume that the voltage
distribution is solely dependent on the capacitances inclu'ded in the equivalent circuit
in Fig. 14-2b (for the time being, we ignore the part of the
circuit shown by dashed lines).
Let us replace the lumpedcapacitance circuit in Fig .
.0 x
14-2b, composed of n elements,
1.0 0.8 ' 0.6 O. lf 0.2 0
:": "
' by a distributed-capacitance
circuit shown in Fig. 14-3.
. '~R:
A A)------9X
Then the total series capaciI
tance will be
..( dQ.x '"

'" .




' ,I

Fig. 14-3 Voltage distribution

in the -capacitive circuit of a
winding -with its finish grounded:
' . , . .. ., . . . .
i-initial distribution at a
. = 10; 2':"""initial distribution at
a = 5; 3-initial distribution
in a transformer with electrostatic shields; 4-initial distribution at a = 0 and final distribution at any value of a

c, =

1/LJ (1/Cd) = Cdln

and the total shunt capacitance

will be

c, =.LJ
Cq = c;,


Taking the winding length as

unity, we may, for a winding
element of length dx, find its
shunt capacitance C q dx and
the shunt differential parameter K dx, where K = 1/C dWith V m at the start of the winding held constant, the
voltage v'" within a distance x of the grounded terminal X
can be found by solving a set of differential equations for
the shunt charge on the element K dx, equal to

Q", = dv",/K dx


and the voltage across the capacitance C q dz, equal to

Ch.14 Overvoltage Transients in Transformers


On finding the derivative dQx/dx from Eq. (14-1) and

substituting it in Eq. (14-2), we obtain a linear differential
equation with constant coefficients for V x :
d 2vx/dx 2 - (C q/Cd) V x = 0

Voltage Distribution over

the Winding and Its Equalization

The solution of Eq. (14-3) has the form

V x = D 1 exp (ax)
D 2 exp (-ax)

where a = 11 C q/Cd and -a are the roots of the characteristic equation. Applying the boundary conditions that exist
when the winding terminal is grounded
,'(i ) V x = D 1 exp (a)
D 2 exp (-a) = V m for x = 1
~ (2) V x -:. D 1
D z . 0 for x = 0
we can find the constants D 1 and D 2' and the initial voltage
distribution will be
V x = V m sinh ax/sinh a
As is seen from Fig. 14-3, for the values of a most frequently found in practice, a = 5 to 10 (curves 1 and 2),
the initial voltage distribution is rather nonuniform, and
becomes the more so as a increases (that is, with an increase
in the shunt capacitance or a decrease in the series capacitance).
When the initial voltage distribution is ideally uniform
(curve 4) corresponding to a ~ 0, and
, '
V x = V m sinh ax/sinh a ~ V max/a = V mX,
the -voltage existing across the element I1x nearest to the
start of the winding is
I1v = V m I1x
, In a real winding (a ~ 3), the voltage existing across
the winding element I1x nearest to the start [see Eq. (14-4)]
will be
I1v = (dvx/dx)=l I1x = (V ma coth a) I1x
. Vma Ax
which is a times the voltage in the case of a uniform distribution. .


Pa rt One. Tran sformer s

.The fur th er propaga tio n of the over voltage wa ve along the

transformer winding can conveniently be examined, if we
assum e that the wave is rect angular in sha pe , as shown by
the dash ed line a t. the bottom of Fig. 14-1. In this case,
V m at the winding start, appearing a t. t; ~ 0, remains
unchanged, and soon all poin ts on the winding come by a
steady-stale voltage. This is th e final voltage distribution.
With the win ding t erm inal ground ed, this will be a linear
V x = V mx'
(curve 4 in Fig. 14-3).
The propagation of an overvoltage wave along the winding
is in effect a transition from the initial voltage distribution
at t = t; ~ 0 to the final voltage dis tribution at t = oo ,
Because the equivalent circuit of the winding is composed
of capacitances and inductances which form , between them,
a cascade of resonant circuits, the transition from the initial
to the final voltage distribution at each point is oscillatory.
Owing to the losses in the ' resistanc es, these oscillations
gradually decay. The swing of oscillations and the associated
overvoltage inc rease wi th increasing diffe rence between
the initial and final voltage distributions.
To minimize the hazards associated with such oscillations,
the value of a mu st be kept as small as practicable. Also,
a decrease in a leads to a decrease in the initial voltages
existing across the elements close to the start of the winding.
Unfortunately, an ample spacing between the winding and
the grounded par ts cannot be obtained without a marked
increase in the size and cost of the transformer.
The best way to equali ze the initial voltage distribution
and to mak e it comparable with the final distribution is to
use electrostatic shields in the form of open metal rings
(labelled 3 through 6 and shown dashed in Fig. 14-2a). When
such shields are connected to the start of the winding, the
capacitive coupling of the first coils to the winding start
(via capacitances C~h' C~h, and C~;/ in Fig. 14-2b) is sub stantially inc reased, and the initial voltage distribution
becomes more uniform and close to the final one (curve 3
in Fig. 14-3).
In Soviet-made transformers, the use of shields around
the windings gu arantees the required pulse strength of insul ation in transformers for 110 kV an d higher (see Table

Ch, 15 Special-Purpose Transformers

* 15

Special-Purpose Transformers




This chapter discusses special-pur pose transformers which

either transform or convert some parameter(s) of electric
energy (frequency, number of phases, or voltage waveform)
or serve some special purposes (continuous voltage adjustmen t, supply of high voltages, isolation of the secondary
current from the load impedance, supply of secondary current
or voltage proportional to the primary one , et c.).

Three-Phase Transformation with Two


There are ways to transform three-phase with only two

transformers. One of the most commonly used conne ctions
for this purpose, known as
the Scott transformer or the
T-connection, is shown in
Fig. 15-1.
In the T-connection, one
transformer, b, has its primary
connected directly acros s two
lines; t his is t he "main" transWz
former ha ving W I turns and a
primary voltage equal to the
line voltage V C B = VI, line'
The voltages V c o and V o B
are equal, becaus e the emfs
in the two hal ves of t he Be
winding are induced by t he
same magnetic flux (Db' The
other transformer, a, is called the teaser. Its primary has
~~~~i~: -l Three/two-phase con- V 3' wl /2 t urns and is connec te d
between t erm inals A and 0
of the main transformer, so t hat its primary voltage is
V3V I ,li n e / 2 . The second ary voltages Va and Vb form a
balanced two-phase sys tem, becau se they are equal in
12-0 16 9


PArt One. '1'rnnsrormerA

11b= 11 C B (10 2 /10 1 ) = VI, I it le (10 2/W 1 )
Va = llAO (210:/ V 3 WI) = Vi. line (10 2110 1)
and are shifted in phase by the same angle as 11AO and V C B,
that is, 'Jt/2.

Frequency-Conversion Transformers

(a) Frequency trebling. This purpose can be served by a

bank of three single-phase transformers whose primaries are
star-connected and energized from a three-phase supply at
freq 8ncy t.. As has been shown in Sec. 7-3, the fluxes of





Ftg, 15-2 Frequency doubling

such a transformer contain a sizeable proportion of the

third harmonics for which f3 = 3fl' The third-harmonic
emfs, E 3 , induced into the secondary are in phase, so when
the output winding is a combination of three single-phase
secondaries connected in series aiding, the output will only
contain 3E 3 at 3f1' but no emf at the fundamental frequency,
because in a balanced system the fundamental emfs sum
to zero.
(b) Frequency doubling. This purpose can be served by a
transformer having two independent magnetic circuits
(at o: and ~ in Fig. 15-2). The primary energized from a
supply at t, encloses both magnetic circuits, so the emf
induced in it is due to the sum of two fluxes, (I)a and <Dfl.

eh. 10 Speelal-Purpose Transformers


The secondary in whi ch the emf is induced at twice the

fundamen tal fr equ enc y , I'!. = 2/1' has lo'!. t urns contributed
by two halves whi ch are wound on diff eren t cores and are
conne ct ed in opposition. Wi th this arrangement, the flux
linkage of the secondary is proportional to the difference
flux, cD a - cD 13 ' As is seen from the plot in Fig . '15-2, as
the sum flux cI) a
cD 13 al ternates a t frequency 11' t he
difference fl ux o, - <Dj3 altern ates at twice that frequency,
I'!. = 2/1' if the two cores are biased in opposite senses by
winding 0 energized with direct current I o an d establishing
an mmf F 0 = lOoI o in eac h core .


Variable-Voltag e Transformers

Stepwise voltage adjust men t by tap cha nging has been

examined in Chap . 6. Because of an added comp lexity in
transformer design and t ap-changers, only several steps of
ad justment can be provi ded,
a ?+~
and t he overall ra nge of adj ustment does not exceed +5 %.
A more continuous voltage
adjustmen t in low-volt age
transformers and low-power
f::b.. v~l
au t otransformers can be obtained with brushes or sliding
contacts that can be mo ved
acr oss a skinned portion of
he transformer winding. This
Iform of adjustment changes
1 f=
the transformer voltage in
0'-steps equal to the voltage
11 11
one t urn ; the range of
adjustmen t can be ex tended
Fig . 15-3 Vari abl e transformer
considerably. This arrangement
with a d.c.-biased core
is utilized , for example, in dim mer-control transformers for
auditoriums and theatre stages. In a 250 kVA unit, the
output voltage can be varied from zero to 220/380 V.
In high-power or high-voltage t r ansformers, cont inuous
voltage adjustment can be obtained by biasing the core
with direct current. A li kel y arrangem ent utilizing d.c. bias
is shown in Fig. '15-3. It is a combination of two single-phase
transformers differing in the transformation ratio, n'!.la




Part One. transformers


nZlll' Each transformer has a split core biased with d.c,

ill the same manner as ill the case of frequency doubling.
(When the core halves are biased in opposite senses, the
biasing current is nearly sinusoidal.)
The primaries with voltages VIa anrl Vl ll are series-con.
nected for a supply voltage VI = VIa
Vll~ ' The secondaries with voltages -Vza = nZlaVla and -V ZIl = nZlllVl ll
are likewise series-connected and loaded into an impedance
the voltage across wh ich is

Vz = V za

+ V ZIl

When the cores of transformers a an d p are biased separate ly , it is possible to vary the ratio between the resistances
of the primary windings traversed by a common current II,
and the voltage ratio


= VlllIV l a

For example, if we increase the bias on transformer p, the

voltage ratio will decrease. A change in 1; brings about a
proportionate change in t he output _yoltage



2 -

V2a _ V2/3 ---=

V1 n21a'1+,n21/3
+ ,

When 1; = 0, - V z = Vln Zl a; when 1; = 00, - V z = VlnZlll.

In practice, the output voltage can be varied within narrower, but sufficiently broad limits.

Arc Welding Transformers

Arc welding transformers ha ve to operate intermittently,

with frequent transitions from no-load to an arc often accompanied by instantaneous short-circuits.
It is usually required that the short-circuit current of a
welding transformer be not more than two or three times
its rated voltage. Another requirement is that variations
in the circuit (load) impedance ought not to produce marked
variations in output voltage. To meet these requirements,
the short-circuit impedance of a welding transformer must
be many times that existing in ordinary transformers. As a
rule, the short-circuit impedance of welding transformers is
raised at the expense of the inductive reactance . To this end,

Ch. 15 Special-Purpose Transformers


the windings are placed on different sections of the core and

are series-connected . A further increase in short-circuit
inductance can be obtained by pl acing adjustable reactors
in the secondary circuit.

Insulation Testing Transformers

Insulation tes ting uses voltages from 1 MV upwards. Such

voltages can only be supplied by a cascade of series-connected
transformers (Fig . 15-4). The to tal output voltage V is the

Fig. 15-4 Three-stage casca ded l aborat ory transform er:

(a) circuit diagram; (b) external appearance of a 1.5-MV, '1.5 MVA
cascaded transformer (in contrast to the arrangement in (a), the trans-

former tank in the firs t st age is isolat ed fro m g roun d)

sum of the second ar y voltages, V 2 , supplied by each st age

in a cascade. In a three-stage cascade, it is V = 3V 2 Each
transformer in the cascade is installed in a separate tank
and has three windings, namely winding 1 energized from
the previous stage, and windings 2 and 3 which are autoconnected (the last unit in the cascade has only Winding
1 and 2). The tanks of the second and third stages stand on
pedest al insul ato rs and are at a voltage of V 2 and 2V 2
relative to the ground, respectively. The tank of the first

Part One. Transformers


stage is grounded . Accordingly, the winding insulation of the

first and second transformers] is designed for V 2
V 3'
and that of the third, for V 2


Peaking Transformers

Peaking transformers are employed in electronics as sources

of recurring peaked voltage pulses which have a short
duration in comparison with the pulse period. Such a voltage can be produced across the secondary of a transformer
with a heavily saturated core , if the primary is connected to
a source of sinusoidal voltage, VI = VIm sin tat, via a highvalue resistance R or a linear inductive reactance. Then
the primary current i l will be sinusoidal, the flux waveform
will be flattened, and the emf will be peaked. A similar
effect, although less pronounced, can be obtained with a
star-star three-phase transformer (see Sec. 4-4 and Fig. 4-14).

Instrument Transformers

Most power systems operate at voltages and currents too

high to be measured by ordinary instruments directly. Instead, instruments must be connected to an H.V . network
via instrument (or measuring) transformers which may be
designed for voltage or current measurements. Instrument
transformers are also used to energize control and safety
relays and other automatic control devices. Instrument load
on the secondary of a measuring transformer is called its
burden and is expressed in volt-amperes at a certain power
Voltage transformers. Instrument transformers in this
group are designed to step down the primary voltage to
around 100 V. The burden should be no Iess' than some specified value, ZR, and the transformer must be designed so
that its referred secondary voltage changes little as the
load is varied from zero to its full (rated) value.
In Sec. 5-1 it has been shown that

v: ~ - T\Z' /(Zsc + Z')

Therefore, if Z' ~ Z sc, then V~ ~ -VI. When this condition
is met, the primary to secondary voltage ratio will always

Ch,-15 Special-P urpose Transformers


be the same, and


VI W 2 / W 1

V1n 2 1

For all measures taken , Zsc will alw ays be greater than
zero , and therefore a voltage tra nsformer in troduces two
kinds of error in the measuremen ts being made: the ratio
error and t he phase-angle error. The ratio error is given by
, t v = V 2 lV / W 2 - VI X 100%






an d the phase-angle error 6v refers to the phase angle between

Fig. 15-5 Ra tio 'and pha se-angV j and - V~(see Fig. 15-5a).
le errors of voltage and curre nt
Th ese errors increase with
instrument transformers
increasing Z and ought not to
exceed cer ta in limits specified
in appropriat e standards. The limits of error define the
"accuracy class" of a transformer and are state d for the rated
burden Z = ZR and t he rated primary voltage. The error
limits for three accuracy classes adopted in the USSR are
given below.

Accuracy cla ss 1: t v

= +0.5 %, 6 v = +20' ,

Accuracy class 2: t v = +1.0% ,

Accuracy class 3:

6 v = +40'

tv = + ~.0 %, 6v

, no .liin~t

Current transformers. Instrument transformers, in this

group ar e intend ed to change curren ts in power networks ~Q
values acceptable :.to' in~ters , usually' down t o 5 A. ' ,' -::',',
As 'already noted ;' the' secondary of a currentfransforfner
is connected t o, 'an ammeter, a wattmeter, or an automatic':
cont rol device. if several in struments ar e powered 'by: the
Same current transformer, they 'are series-conn ected. .
' For propel' operation, a current transfor iner ' must: he
heldin astate close to a short-circuit (Fig. 15-5b)'. Its 'burden
Z ',ought not to exceed a certain rated value, ZR. As follows
from "t he basic equations and equivalent circuits' of transformers (see Sec. 3-5) , the secondary current is related to
th'E{primary ' current by 'the following equation: ' , " ,


j~ = j2W2/WI =-i1zO/(Zo

+ Z~ + Z')


' "


I:'~ ~

' :': !

Part One. Tr ansformers


It is an easy matter to see that the ratio error and the phaseZ'
angle error will progressi vely decr ease as t he sum Z~
decreases in com paris on with ZOo This is why t he designer
makes every effort to t urn out a curre nt t ransformer having
the highest possible value of Zo, the lowest possible value of
Z~ , and with Z < ZR'
At the rated burden, the curre nt ra t io errors '

t, = 12W2/~ 1-Il


an d t he ph ase-angle error 6; (see Fig. 15-5b) should not

exceed t he limits stated in applicable standards. In the
Sovi et Un ion , t he following error limits and accuracy classes
are adopted for current transformers.
Accuracy cl ass 0.2: tv = 0.2 %, 8 i = 10'
. Accuracy class 0.5: tv = 0.5 %, 8 i = 40'
. ' Accuracy cla ss 1:
tv = 1.0%, 8i = 80'
Accuracy class 3:
tv = 3. 0% , 8 i = no limit
Accuracy cla ss 10: tv = 10 %, 6 i = no limit


Heating and Cooling

of Transformers


Temperature Limits for Transformer

Parts under Steady-State and Transient

Energy conversion by transformers involves a loss of power.

The magnitude of power loss varies with the conditions
under which a t ransformer is operating (see Sec. 6-3). The
bulk of t he power lost is dissipated as heat in the core and
coils. The core loss ma y wi th sufficient accuracy be deemed
proportional to the primary voltage squared, and the copper
loss to the primary and secondary currents squared. A change
in load mostly affects the copp er loss, whereas the core
loss, given a constant primary voltage, remains nearly
,unaffect ed.
Some of the power loss is dissipated also in the structural
parts (the t ank , clamping arrangement, etc.) lying within
the magnetic fi~ld of the transformer.

Ch, 16 Heating and Cooling of Tran sformers


The heat dissipates in the transformer parts, and they rise

in temperature above the surroundings. As the t ransformer
keeps rising in t emp erature, a progressiv ely la rger amount
of heat is transferred to the surroundings , because the hea t
flux is proportional to the temperature rise (the degrees
above the ambient temperature). Given a sufficiently long
time (theoretically , an infinitely long t ime), t he temperature will cease rising, because all of the heat dissip ated will
be transferred to the surroundings . (In more detail, heating
and cooling is discussed in Sec. 35-3.)
The steady-state t emperature of the t ra nsformer parts
depends on the cooling arrangement used.
A transformer an d its cooli ng system must be~designed
so that the temperature rise does not exceed the limit specified in each particular case. The limits given in relevant
standards most apply t o the parts coming in contact with
the insulation, oil or any other dielectric liquid that ma y be
The reason for this is th at elevated temperatures lead to
an accel erated ageing' of insulating materials with the resultant loss of electrical and mechanical strength. E xperiments
have shown that an increase of 8 degrees C in t emperatur e
will halve the service life of an oil -immersed t ransformer .
A transformer will serve reliably for 15 t o 20 years, provid ed the temperature rise of its par ts does not exceed the
limits stated below (the figures are taken from an applicable
Soviet standard).
Oil-I mrnersed Tran s] 0 rmers

Wi ndi ngs
65 dog. r.
Exterior sur faces of core
and stru ct ura l work
75 deg. C
Top layer of oil:
in totally encl osed uni ts
60 deg. C
in oth er types of enc losu r e 55 d eg. C
Dr y Tra nsj ormers

Wind in gs 'a nd core surfaces in

cont act with insulation accor ding to in sul a t ion class:
Clas s Y
50 deg.
Clas s A
65 deg.
Cla ss E
80 deg.
Cla ss B
90 dog.
Class F
115 deg.
Class H
140 'dsg.



Part One. Tra nsformers

The limits of temperature rise stated above are fixed,

assuming that the ambient temperature is 40C. In the case
of water-cooled transformers, the inlet temperature of cooling
water is assumed to be 25C, and the respective limits of
temperature rise may be raised by 15 deg. C.
The design temperatures of transformer parts are assumed
to ensure a service life of 15 to 20 years, in view of the
observed daily and yearly variations in ambient temperature
and transformer load 'under actual service conditions. Most
of the time, the load is less than rated and the amb ient
temperature is lower than 40C, so the transformer insulation
reaches its design temperature but seldom, and this extends
the service life of transformers .
The limits established for the winding temperature under
steady-state short-circuit conditions are as follo ws.
Oil -Immersed Transf ormers

Copper wi nd ings
Aluminium windings
Dry Trans formers

Copper windings and insu lation of the classes

1isted below:
Class A
Class E
Classes B, F, H
Aluminium windings and insula Lion of the
classes listed below:
Class A
Classes E , B, F, H

The short-circuit duration (see Sec. 13-2) must be li mit ed

so that the temperature limits stated above could not be
exceeded . In Soviet practice, it is under 5 s.

Transformer Coo ling Systems

Small transformers are air-cooled and insul ated, which is

why they are usually referred to as dry transformers. For
units of larger rating and higher voltage, oil cooling is
more economical.
Oil cooling may be natural or forced. In the former case ,
transformers are referred to as oil-immersed air-cooled.
The core and coil assembly of such a transforme r is encl osed
in a tank filled with transformer oil. H eat dissipated by the
coils and core is transferred t o the filling oil. The hot oil

Ch. 16 Heating and Cooling of Transformers


is lighter than the cold oil next to the tank sides, and this
difference gives rise to a natural circulation of oil in the
tank. On picking up heat from
the hot parts, it rises; near
the tank sides, it gives up its
heat and sinks. The heat transferred to the tank sides is then
reje cted to the surrounding
air (Fig . '16-1).
Under steady-state cond it ions, t he temperature distribution in each h ori zontal
layer is such (Fig . 16-2) that
100 r . the temperature rise of the
core and coils relative to the
Fig . 16-1 Variations in temp eoil, on the one hand, and the
rature wi th height of a transformer :
temperature rise of the oil
l-oil temperature; 2-tank side
relative to the surrounding
tem perature; 3-winding temair, on the other, is sufficient
perature; 4-core temperature
for all the heat dissipated by
the core and coils to be transl V winding
by convection to the oil
HY windin!l
the oil to the tank
Tank side
sides, and by convection and
radiation to the ambient air .
As is seen from the figure, the
temperature wi thin the core,
coi ls and tank side changes
but little because they are fab ricated
of metals having a high
thermal conductivity . The
temperature change is more
marked in the coil insulation
and also when heat is transferred from the core and the outer
surfaces of coil insul at ion to
/fo L..
oil and from oil to the tank
Fig. 16-2 Horizontal temperasides . The temperature graditure distribution in an oil -iment is especially pronounced
mersed, natural air-cooled transbetween the outer surface of
the tank and the ambient air.
In transformers of hi gh power ratings, the withdrawal of
heat fr om t he tank sides is a problem calling for special


Part One. Transformers

treatment. The point is that the heat dissipated in a transformer per unit area increases in proportion to the linear
dimension . In simpler words, an increased cooling surface
is necessary. This extra surface may be obtained by making
ducts in the core and coils,
providing fins and corrugations
on the tank sides or, which
is the most common method,
using a tubular radiator (Fig.
16-3). In oil-immersed, aircooled transformers, oil circulates through the radiator (s)
naturally, by convection.
In air-insulated, natural-aircooled transformers, the core
and coil assembly is in direct
contact with the ambient air
and heat is abstracted by
Fig. 16-3 Radiator
convection; some heat is withdrawn by radiation.
Large transformers use natural oil circulation and ail'
blast. In them, by directing an air blast onto an ordinary

Fig. 16-4 Radiator blowers

tubular tank or onto separate radiators , the rate of heat

dissipation can be increased several-fold (Fig. 16-4).
Better heat withdrawal is obtained by a combination of
forced oil circulation and air blast. A still better arrange-

Ch. 17 Transform ers of Soviet Manufacture


ment, especially for very large transformers, is to combine

oil immersion and water cooling. In fact, two arrangements
are possible in this case, namely natur al oil circulat ion and
water, and force d oil circulation and water . In the former
case, an intern al cooler is employed, whereas in the latter,
the oil /wa ter heat exchangers are external to the transformers. Oil-immersed , wa ter-cooled tr ansformers require large
amounts of running water, so t hey are mostly installed at
hydraulic power st ations ,


Transformers of Soviet Manufacture


USSR State Standards Covering


In his studies or work, the reader ma y run into transformers of Soviet manufacture. If so, it will be useful, as we
believe, for him t o know whi ch USSR state st andards,
GOSTs, are applicable to various transformers.
GOST 16110-70. Power t ra nsformers. Terms and definit ions (see also CMEA* Standard 1103-78) .
GOST 11677-75. Power t ransformers. Gen eral specifications
(see also CMEA Standard 1102-78).
GOST 721-77, GOST 21128-75. Rated ph ase-to-phase voltages .
GOST 18619-73. Power tra nsformers, three-phase, natural
air cooled, general-purpose, '10 to 160 kVA, up t o 660 V.
GOST 14074-76. Power transformers, dry, protected , general-purpose, 160 to 1.6 MVA, 6 to 15.75 kV inclusive.
GOST 12022-76. Pow er t ransformers, three-phase, oilimmersed , general-purpose , 25 to 630 kVA , 35 kV inclusive .
GOST 12965-74. Power t ransformers, three-phase , oilimmersed, general-purpose, '110 kV .
GOST 17546-72. Transformers (and autotransform ers),
three-phase, power , oil-immersed , general-purpose, '150 kV .
GOST 15957-70. Transformers (and autotransformers),
power, oil -immersed, gener al-purpose, 220 kV .
GOST 17545-72. Tr ansformers (and autotransformers),
power , oil-immersed, general-pur pose, 330 kV.

* CM EA stands for th e Coun cil of Mu tual Economic Assistance

of which th e USSR is a memb er.-Trunslator 's not e.



One. Trnnsformers

GOST 17544-72. Transformers (and auto transform ers) ,

power, oil-immersed, general-purpose, 500 kV .
GOST 3484-77. Pow er tr ansform ers. Test pro cedu res.
GOST '15'16 :1-76. A.C. electrical equipment for 3 to
500 kV. Insul a tion streng th requirements .
GOST '15'16 .2-76. A.C . electrica l equipment and installation for 3 kV and higher . General proc edures for insulation
t esting.
GOST '1 4209-69. Tr ansfo rmers (and autotransform ers) ,
power , oil-immersed . Load capac ity .

Type Designat ions of Soviet-made



typ e design ation of a t ra nsformer consists of letters

numerals. The letters are used as follows .
stands for an autotrans form er .
stands for three-phase . A second T, for three-winding.
o s tands for single-phase .
P stands for a split LV wind ing (see Sec. '10-'1).
H st ands for on-l oad t ap changing. (If there is no H in
the t ype designation, t he transf orm er is designed for offload tap changing or has no t ap chang er at all. )
The numerals in the numerator, following the let ter(s),
give the power rating in kVA, an d the numerals in the denominator, its kV class on the HV sid e.
The designations used for the various cooling arrangements are listed in the t able t hat follows.
Tablc 17-1 Des tgna tion of Cooling
Arrangem en ts
Dry trausformers
Natural air cool ed, open
Same, pr ot ected
Same , seale d
Air-blas t cool ed
Oil- Immersed 'I'ransform crs
Oil natural
Oil -na tural , air-blast
For ced-oil , ai r-bl as t
Oil-natural, wate r
Forc ed-oi I, wa t el'









11 transformers of Soviet. Mnnufncture



Some of Transformer Applications

Dry transformers are mainly intended for installation in

dry indoor locations with a relative humidity of not over
80 % and in the absence of corrosive substances and currentconducting dust. TIleY are fire -safe and are gaining popularity in residential buildings, in laboratories, etc. A dry
transformer may be built into an enclosure so as to kee p
fore ign objects from finding their way into the core and
coil assembly, but to give free access for cooling ail'.
Low-power transformers (under 4 kVA for single-phase
uni ts and under 5 kVA for three-phase units) find use in
radio, electronics, automatic control, communications, in dustrial drive, domestic appliances, and to energize handheld power t ools.
All transformers are designed for moderate climates, for
tropical climates (tropicalized), and for cold climates.

A General T heory
of Electromechanical Energy
by Electrical Machines


Electromechanical Processes
in Electrical Machines*


Classification of Electrical

An electrical machine operating by electromagnetic in duction consists essentia lly of a station ary member and a
movable memb er (Figs . 18-1-18-5) .
The st ationary part is m ad e up of a suitably shaped core,
one or more windings, and structural parts intended to hold
the stator in its designated position .
The movable part consists of a core , one or more windings,
and structural parts ena bling t he m ovabl e part to move
relative to the stationary part and to tran smit mechanical
energy to or from the m achi ne .
The movable and stationary windings may be connected
to external lines directly or thr ough a suitable device.
Conn ection to the movable windings is by sliding contacts.
As a rule, the mo vable part of an electrical ma chine has one
degree of fre edom (motion in an y other directions is prevented by bearings or supports whi ch may be of one of several
In most electrical machines, t he mo vable member rot ates
relative to t he stationary member. Quite aptly, they are
called rotating machines, and their movable member is
called the rotor, and t he stationary member t he stator.

* Th e auth or refers primarily to th e moto r mo de of operation.

By th e reversibility pr in ciple, however , th e rea der m ay readily extend th e reasoni ng to th e generati ng mode where necessar y.-Translater 's no te .

Fig. 18-1 Ro tatin g cyli ndrica l ma chine:

I - stator wind ings; 2- rotor windings; 3-st a tor core; 4-rotor core;
5- stat or st r uctura l parts ; 6-rotor sha ft ; 7-axia l-radia l bearin gs

Fig. 18-2 Rotating diso-t ypej machine (the notation is the same as:in
Fi g. IS-l)






Fig . 18-3 Flat lin ear machine:

I -stator windings; 2-movable-ll1ell1ber windings; 3-stator core;
4-movable-mell1ber core; 5-stator structural parts; 6-movablemember connec ting-rod ; 7-supports
I 3- Ut li 9

Part Two. En ergy Conversion by Elect rical Machines

Most fr equently, the roto r is a cylinder rot ating ins ide

the stato r which is likewise a cylinder bu t a holl ow one
(see Fig. 18-1). Somet imes, to incre ase the moment of inertia
of the rot at ing parts , the rot or is made in t he form of a ring
enclosing the stator.
As an alt ernative, a rotating machi ne can be built so
that the stator and the rotor are discs facing each oti er
(Fig . 18-2).
A less frequent variety of electrica l machi nes is one in
which the movable part recipr ocates rel ative to the stator

Fig . 18-4 Tubular linear machi ne (the notation as in Fig. '18.3)

in a linear fashion. Quite aptly, such machines ar e calle d

linear . They ma y be flat and t ubu l ar .
In a fl at linear m achine, the movable an d t he stationary cores are each the shape of a parallelepiped , with their
broad sides facing each other (Fig. 18-3). In a t u bu lar
linear machine, the movable cylindrical core is free t o move
axiall y inside the stationar y annular core (for exa mple, a
plunger moving inside a solenoid), as in Fig. 18-4. Flat
linear machines can serve as, say, drives for electric-powered
rail-riding vehicles, especially where high speeds (over 200
or 300 kmph) are in vol ved or desired. Tubular linear
machines can be used to actuate the reciprocating parts of
various mechanisms.
Both rotating and linear machines can be built for restricted to- and-fro motion . Restricted rotary motion may
be utilized t o operate, for example, the balance wheel of
an electr ic clock , and restricted linear motion ma y ser ve
t o actuate an electric pick.


tho 18 Proces ses in Electrical Machines

Sometimes, it may be necessary to link an electrical

machine to a source (or sink) of mechanical energy so as to
transform some parameter(s) of the mechanical energy being
converted . Thi.s i.s done by what may be called mechanical


Fig. 18-5 Geared rotating electrical machine:
I-frame; 2- st at or core; 3-stator winding; 4-rotor; 5-rotor shaft ;
6- ball bearing; 7-gear train; 8-gear-train shaft

converters. A mechanical converter is often made integral

with the associated electrical machine. The most commonly
used form of mechanical converter is a step-up or a stepdown gear box (Fig. 18-5).
Rotating motion can be transformed to reciprocating
motion by gears, a worm and gear combination, or friction
transmission. Osolllatory motion can be transformed into
rotating or translational motion by a variety of ratchets and
Most frequently, however, electrical machines are built
without any mechanical converters .

Mathematical Description
of Electromechanical Energy
Conversion by Electrical Machines

Let us consider a rotating electrical machine in which the

windings have an arbitrary number s of parallel paths (or
circuits) embedded in slots or on the outer surface of the
stator and rotor. Each path may consist of many coils
connec ted in some particular manner. The cores, too, may be


Part Two. Energy Conversion by Electrical Machines

of any configuration. As an example, Fig. 18-6 shows a rotating electrical machine with s = 5 parallel paths of which
two (labelled "1" and "2") are located on the stator, and three
(labelled "3", "4", and "5"), on the rotor.
The electromagnetic processes taking place in an electrical
machine can be described by Kirchhoff's mesh (or loop)

Fig. 18-6 Multiwinding rotating electri cal ma chine

equations and the equations of motion for the rotor. In a

linear approximation (that is , assuming that the core material has an infinitely large permeability) , the flux linkage
of, say, the kth path (where k may take on any value from
1 to s) can be expressed in terms of the winding currents im
the self-inductance of the kth winding, L k 1tl and the mutual
inductances between the !dh winding and all the other
windings, L k n , where n can take on any value from n = 1
to n = s, except n = k

\f" =



\f kn = ):


n= l

In many cases, the mutual and self inductances of the windings are markedly affected by whether or not the cores
have saliencies. With saliencies, rotation of the rotor causes
variations not only in the mutual inductance between the
paths on the stator and rotor, but also in the mutual inductances between the paths on only the stator and on only the
rotor, and in the self-inductance of each path.

Ch, 18 P rocesses in Electrical Machines


In the general case, all t he self-induct an ces L k h an d all

t he mutual induc tances L im are functions of the coil and
core size and of the angul ar position of the rot or , L k n = f (1').
Using Kirchhoff's voltage law , we may write a set of s
volt age equations, each describing one of the parallel pa ths.
For t he kth pa th, such an equation takes t he form
Vh = R hi h + (f'P'h/ dt

wher e
L im din/d t

R hi h +



(L hn di n/dt + inQ dLhn/d 'V)


t he transformer emf rela t ed t o var iat ions in

the cur rent in the nth path
in dL lm/ dt = the rotational emf relate d to vari ations in
t he mu tual inductance with the nth pa th
(when n ==1= k) or to variations in the self-inductance of t he ktb path (when n = k)
Thus, as follows from Eq. (18-1), the emfs induced in the
kth loop are the sum of transformer emjs related to vari ations
in coil currents when the mu tual or self inductances remain

- nI:=1

Lim din/dt

and rotational emjs relat ed to variations in the mu tual or

self-inductan ces, with t he currents held constant


2J in dLkn/d'l'
n= 1

The term "transformer emf" refers to the fact that a similar

emf is induced in transformers where the primary and
secondary are stationary relative to each other. The term
"rotational emf" refers to the fact that it can only be generated when the rotor is moving at some angular velocit y
Q = dl'/ dt
For t he loop s connected t o an extern al circuit, Vh in Eq .
(18-1) can be in t erpreted as t he emf of t he circuit . For shortcircuited loops , Vh = O.
The mechanical power derived by an elec trical machine
from electrical energy can be expressed in t erms of the associated circuit parameters , proceeding fr om the law of
conservat ion of ener~y. To begin with, let us determine thy

Par t Two. Energy Conversion by Electrical Machin es


instantaneous electric power that the kth loop draws from

the associated supply line



V h~ h =


~ L
hn crt

+r\~ ' ~'h


~. dLh1l
LJ ~n ----cry-11= 1

The total instantaneous electric power drawn by all the loops

can be found by adding together the powers of all the loops:


Ph =

1<= 1

2j R/,i.~ + 2j
h= 1


h= 1


L hn



el L h1l



The t erm ~ R hi'f, is the power dissipated as heat in the


loop resistances R h and gives the power los t on conversion.

Th e rem ainder of the input power goes to sustain variations
in t he field energy owing to variat ions in t he loop currents
and in duc t ances . Because t he magnetic field energy is


i h 2j inL hn
its t ot al change over a t ime dt during which i/o in , and L h1l
change by di h, d in, an d dL /m , is given by


dfiV = (oW/iJih) di h


h= 1

+ (o W lOi n) di n + (8W/aLhn) dL hn


L/m di n


n =l

i h ~ i n dL hn
n =l

Th erefore , the power spent t o sustain variations in t he

magnetic field energy is

Pw =

~ = L;


n =l



+ (Q/2) ~






It corre spo nds to t he second and half the third te rm in E q.
(18-2). In other words, t he power sp ent to sustain variations
in the energy of t he magnet ic field is all of the sum of t he
powers defined as the pro du cts of t he loo p currents by the

Ch. 18 Pro cesses in Ele ctrical Machines


transformer emf, and half the sum of the powers defined as

the products of the loop currents by the rotational emf.
The remainder is the mechanical power transmitted by
the shaft t o the driven machine (in motoring) or from a prime
mov er (in gener ating)

Pme ch


LJ Ph- Pe- Pw

~~ ""
. LJ
"" In
. ~
dLk n
LJ lk

k= 1

h= 1



As follows from Eqs . (18-1), (18-2) and (18-5), the mechanical

power is equ al t o half t he sum of the powers defined as the
products of loop currents and the rotational emf. Hence, we
ma y conclude that electromechanical energy conversion
involves only the rotational emf , whereas the transformer
emf does not contribute to this conversion. It is to be noted
that the power spent t o sustain variations in the magn etic
field energy is not wasted irrevocably , bu t sums on the
avera ge t o zero. This is because ie a ro tating electrical
machine all quantities (currents, self-inductances, mutual
inductances, etc.) vary periodically. At the end of a cycle
of alternation , all quantities, including the magnetic field
energy, t ake on the same value they had at the beginning
of t he cycl e, i.e. , W et ) = W(t +T ) ' This implies that variations in the energy of the magnetic field over a cycle, or
period, sum to zero , that is ,

.\ dW =

W ( t+T ) -

H l (t)


During t hat part of a cycl e when t he ma gnetic field energy

builds up (dW> 0), the power Pw given by Eq. (18-4) is
positive (Pw > 0), and the energy required to set up the
magne ti c field is t aken by the loops from t he line. During
the rema ining par t of a period , Pw < 0, and t he energy
st ored by t he m agnetic field is again returned t o the line.
This exchange of energy betw een the machine and the line
goes on in such a manner that the energy drawn from the
line averages over a period t o zero. A measure of t his exchange is wh at is called the reacti ve (or magnetizing) power.
In the case of a single-phase supply an d sinusoidal variations,
this is the maximum instantaneous power drawn from the
line to set up the magnetic field in the machin e;


I dW/dt

Im! 1\

Part Two. Ene rgy Conversion by Electrical Machines


Recalling t h at the electromagnetic t or que, T em, acting

on the rotor at a given instant can be expressed in terms of
Pmech defined in Eq. (18-5), an d comparing the resultant
expression with Eq . (18-3), we get


Pm ech

2 LJ

i " i el Llm
el y

1< =1


(il< = cons ta nt)

n= 1

Thus, in a machine with a linear magnetic circuit t he
electromagnetic t orque is the partial derivativ e of the
magnetic field energy W with respect to t he angular pos ition
y of the ro tor , with t he loop currents held constant (i h =
= cons tant , and in = constant). If this derivative is positive, t he torque act s in t he direction of ro tation (or in t he
direction of increasing y), and elec tric energy is converted to
mechanical. If the derivative is negati ve, reverse conversion
takes place .
Equation (18-6) may be extended t o .machin es with non linear magnetic circuits, if variations in the magnetic fi eld
energy, dliV, as the rotor turns through an angle dy can be
foun d not onl y for i k = constant , but als o for f!aj = constant . In each jth elemen t of t he ma gn etic circuit f! aj must
be found for i h = constant and the angular posit ion '\' of
the rotor.
If the te rminal coil voltages VI<, the angular velocity Q
of the rotor , and the rela tion Lim = f (y) are known or specified in advan ce, the currents ca n be found from Eqs. (18-1).
J,hen t he electromagnetic torque can be found by Eq . (18-6)
where y = Qt .
If the angula r velocity is not known , hu t the external
torque T ext is specified in advanc e, then Eqs . (18-1) and
(18-6) must be solved simultaneously with the equ ations of
motion (18-7)
T em- T ext

= J c!Q/dt

Q =Q lnit +

I (dQ/ d t) dt

l' =

'\'Inlt -/-


Q elt


Gil. t9 Production of Periodic Magnetic Field


The mathematical description derived above for a multiloop (multipath) rotating machine can be extended to a
linear machine whose movable member reciprocates relative
to the stator.
The equations for a linear machine differ from the above
equations only in that the angular displacement y is replaced
by a linear displacement x , the electromagnetic torque Tern
is replaced by an electromagnetic force N acting in the
direction of displacement, the angular velocity Q by a
linear velocity u, the angular acceleration dQ/dt by a linear
acceleration du /dt, the external torque T ext by an external
force Next ' and the moment of inertia of the rotor J by
the mass of the movable member, m:


Production of a Periodically Varying

Magnetic Field in Electrical Machines


A Necessary Condition for

Electromechanical Energy Conversion

From inspection of Eq. (18-6) , we may conclude that a necessary condition for an electrical machine to perform electromechanical energy conversion is a change in the self or
mutual inductances of the coils as the rotor turns through
an angle. An electrical machine will perform its function if
the derivative of at least one quantity with respect to the
angular position of the rotor is non-zero
dLkn/dy =1= 0
because it is only then that Tern =1= 0 and P mech =1= O.
This is a necessary, but not a sufficient condition for a
continuous, unidirectional electromechanical (or mechanoelectrical) conversion of energy. It is also required that Lhe
currents in coils 11; and n should vary in such a manner that
not only the instantaneous, but also the average values of
T em and P m ech be sufficiently large.
Because in technically feasible designs the magnetic fields,
flux linkages,iself and mutual inductances cannot be monotonically rising functions of currents and the angular position
pi the rotor" the only possible case is when these quantities


Pa rt Two. Ene rgy Conversion b y El ectrical Machines

var y periodically as functions of 1', when the derivative

dL/m/d'V likewise varies periodically.
For L ll n to be a periodic function of 1', it is essential that
t he current tr aversing coil n sets up a magnetic field periodically varying in sp ace (t an gentially t o the air gap).
Some of t he coil an d core desi gns capable of producing a
periodicall y varying magnetic field are discussed in the
sections t hat follow.

The Cylindrical (Drum) Heferopolar

W inding

The conductors of a drum winding are laid in slots on the

side surface of the core which ma y be in the form of a toothed
(or salient-pole) cyli n der or toroid*.
As is seen from Fig. 19-1a, the curre nt in the conductors
on the core surface facing the air gap alternates in direction
periodically. This gives rise to a magnetic field which varies
periodically in space - the core is magnetized heteropolarlyin going r oun d the circumference, an N pole is followed by
an 8 pol e, and an 8 pole is foll owed by an N pole.
The sp acing between zones A an d X occupied by conductors
carrying currents which alternate in the direction of flow
varies from design to design. Accordingly, a drum windin g
can set up a magnetic field wi th a varying number of periods ,
cycles of altern at ions per r evolution , or , as mo re commonly
stat ed, a vary ing number of pole pa irs, (Fig. 19-2).
This spacing is me asured along t he periphery of the airga p with a mean radius R an d is call ed the pole pitch. If we
designate the pole pitch as .. (see Fig. 19-'1a and Fig. '19-2),
t hen the numb er of pol e pair s (or cycles of alternation per
revolution) will be give n by

= nD /2.. = nR h:


The simplest of all drum windings is the two-pole winding

for which p = '1 , and the magnetic field comple tes one cycle
of alternat ion per rev olution . Those with p > 1, are called
multipole windings.

. * In mach ines wi th smooth cores (those havi ng no slo ts), the coil
conductors are bound ed t o t\Ie oute r surface of the core :

Ch. ill Production of Periodic Magnetic Field


In a drum winding, the conductors lying on the surface

of the air gap may be interconnected in anyone of several


~..., ,,,,

lol ~


Fig. 19 -1 Production of a periodically varying magnetic field in rotating electrical machines:

(a) cylindrical (drum) heteropolar winding; (b) toroidal heteropolar
winding; (c) ring winding and claw-shaped core; (d) ring homopolar
winding and toothed core

ways. Whatever the form of connection , however, the coilends will never encircle the yoke of the core .
Each coil may be wound with one or two turns . Each slot
may contain one side (Figs, 1~-1a and 19-2a) 0):' tW9 sides

Fig . 19-2 Heteropolar cyli ndrical (drum) windings :

(a) sin gle-layer, single-phase concentrated winding; (b) two -layer, si ngle-phase concent rated wind in g; (e) two-row, single-phase concent ra ted winding; (d) single-layer, single-phase dist ributed
winding (q = 3); (e) si ngle-layer, two-phase distributed winding (q = 3) _,

Ch. 19 Production of Periodic Magnetic Field


(Fig . 19-2b and c) of a coil. In t he former case, we have a

single-layer winding, and in t he lat ter, a double-layer wi nding
if a given coil has one of its sides at the bottom of a slot
and the other side at the top of the same slot (Fig. 19-2b).
If the sides occupying the same slot li e in the same plane,
we have a double-row wi ndi ng (Fig . '1 9-2c).
Frequently, it is convenient to pl ace con duc t ors carrying
currents flowing in t he same direction in several, say three,
slots (Fig. 19-2d) rather than in one. The number of slots
occupied by a phase belt (that is , by a belt of phase conductors carrying currents flowing in t he same direction)
in a single-layer winding is called the num ber of slots per
pole, denoted by q.
When q = '1, the winding is called concentrated. When
> 1, we have a distri buted winding.
The slots of t he same core may carry sever al identical
heteropolar windings producing between them fields with
the same number of cycles of change, p, and energized from
(or supplying power to) a polyphase line. This structure is
called a polyphase winding. As is seen from Fig. 19-2e which
shows a two -phase winding, each phase is a distributed
heteropolar winding with q = 3 (see Fig. 19-2d). Phase A
consists of belts with conductors carrying the current flowing
in the forward direction (A) an d belts with conductors
carrying the current flowing in t he reverse direction (X),
and the spacing between the adj acen t bel ts is equal to the
pole pitch 'to Phase B consists of belts with conductors
carrying the current in the forward direction (B) and belts
with conductors carrying t he current in t he reverse dir ect ion
(Y) . The bel ts of phase B are laid bet ween the bel ts of phase
A and are displaced from the ph ase A belts through a quarter
of a cycle, or a half pole pitch, 't/2.
A similar arrangement is applicable toa polyphase winding with m phases. In such a case, the number of slo ts per
pole per ph ase, q, is given by

Z I2pm


where Z is the total number of slot s on the core. The adjacent

bel ts in a given phase are displaced from one another by a
pole pitch 't, and the belts in the adj acent phases by a distance equal to -clm:


Part Two. Energy Conversion by Electri cal Machin es

The Toroidal Heteropolar Wi nding

A toroidal winding (see Fig. 19-1b) differs from a cyli ndrical

in that the connections between its conductors carrying
currents in the same direction, that is, the coil end s or
overhangs, are wound around the toroidal core . If the Conductors on the surface facing the air gap are arranged in the
same manner as in a cylindrical coil, a toroidal coil does not
differ from the latter as regards the production of a magneti c
field periodically varying in space . In fac t, it comes in the
same modifications as t he cylindrical winding. It offers some
advantages in t he manufacture of small elec trical machines.

The Ring Winding

and a Claw-shaped Core

80 far we have dealt with forms of winding in which a periodic

heteropolar magnetic field was produced by an alternation
in the direction of current flow in the conductors. In a ringshaped winding (see Fig. 19-1c), a periodic field is obtained
due to an alternation in the direction in which the clawshaped tee th of the core enclose the energized ring-shaped
winding. As regards the production of a periodic field, this
arrangement is equally efficient as the previous designs.
A limitation of this design is an increased magnetic leakage between t he claw-shaped polepieces. An advantage is
simplicity in manufacture. Its application is mainly in small
machines and also in special-purpose medium-power units.


The Homopolar Ring Wi nding

and a Toothed Core

A ring winding whose coils enclose the shaft of an electrical machine produces a homopolar field in the air gap.
For the direction of current flow shown in; Fig. 19-1d,
the surface of the inner core is in N polarity, and that of the
out er , in 8 polarity.
Periodic variations. in the magnetic flux density within
the air gap occur owing to the saliencies made on the core
surface facing the air gap. If the surface of the other core is
smooth or has a slight salience, then within the low areas
(slots) the specific permeance will be smaller than it is
within the saliences (teeth). Accordingly, the magnetic

Gil. 20 Basic Machine Designs


flux density within a salience (tooth) will be higher than it

is within a low area (slot).
The magnetic flux density in the air gap will vary with
a space period equal to the tooth pitch, or spacing between
adjacent teeth, t z . The number of pole pairs will be
p = nD/t z = Z
':. ('19-3)
where Z is the number of teeth in the core.
An advantage of this design is that the resultant magnetic
field undergoes a larger number of alternations per revolution
than with any other design because the coil conductors need
not be laid in slots (the ring winding is external to the core),
and there is no limit to slot size-in fact, they may
be however small.


Basic Machine Designs


Modifications in Design

In the previous chapter, we discussed the ways and means of

producing a periodic magnetic field in an electrical machine.
Now we shall see how an electrical machine must be arranged
for the self and mutual inductances of its windings to be
functions of the angular position of the rotor and to vary
periodically as the rotor rotates.
This effect can be obtained in anyone of three basic machine designs, namely:
(1) in a machine with one winding on the stator and one
winding on the rotor;
(2)- in a machine with one winding on the stator and a
toothed rotor;
(3) in a machine with two windings on the stator and a
toothed rotor.
Each design may come in several modifications. As is
explained in Chap. 19, the magnetic field in the air gap of
an electrical machine may be either heteropolar or homopolar. Respectively, one uses two varieties of windings,
heteropolar and homopolar. Heteropolar windings may be
single-phase and polyphase. Homopolar windings may only
be single-phase, and they may operate on a.c. or d.c. Instead of a single-phase heteropolar winding, use is sometimes made of a ring winding in a claw-shaped core.


Part Two. En ergy Conversion by Electrical Machines

Table 20-1. Conceivable
Wind ing 1 on

Winding 2

Rotor core

H et eropolar
the rotor

To ot hed s tator co re

Toothed rotor core

2 Smoo th rotor core

3 T oothed r otor core

H eteropol a r
th e stator


Sm ooth rotor core

cor e


Uncapahle of
Sim ilar to C3

Homopolar a ll th e
6 Smooth


Uncapab le of

Ch. 20 Basic Machine Designs


Designs of El ec trical Mach i nes

the s t ator
Ho rnopo l ar



Sm ooth st a t or cor e

I Too th ed s t a tor co r e I D 1Smo o t h sta to r co re

Sim ilar to C3 (bu t

needs exter nal l eads)

Uncapab l e of energy
conversion ill ei ther
Sim ilar to D3 (but
needs ex te rnal leads)

energy con versi on ill eit he r d i rec tion

Sim ila r to D:l

energy conversion in eit her d ir ec tion

Uncapable of ele c trom echani cal or mochanoel ec t r ical conversion


Part Two . En ergy Conver sion by Eiectrical Machin es

Winding 1 on
W inding 2

Rotor core

Tooth ed


Smoo th


Toothed st a t or core


Not e. H et er opolar win ding 1 may be

Uncapabl o of
single- or pol ypha se. Ramap o

The core s urfa ce faci ng the air ga p may be smooth or

to othed . Accord ing ly, there may be smo oth cores and t oothed
cores. In toothed cores, the opening and, sometimes, the
shape of slots and teeth have a direct bearing on the permeance of the air gap . In fact, the teeth can be suitably
shaped to control the permeance of the air gap . In smo oth
cores , the slots have a limited opening (as compared with
the air gap) , and the air gap between the face of a t ooth and
the mating core remains constant as the rotor rotates. In
such cores, the opening of slots has a negligible effect on the
perm eance of the air ga p .
Toothed cores are used in electrical machines in which
energy conve rs ion is based on peri odic variations in the
permeance of the air gap . In some cases, the teeth of such
cores are.dimensioned so' as to obtain a desired shape for the
field in the air gap (this is true of the salient-pole roto rs of
synchronous machines and the salient-pole stators of d.c,
machines) . Also , cores in which the slots are made open
for ease of coil placement behave like toothed cores. The
use of open inst ead of semi-closed slots leads to in creased
pu lsational losses and is justified only inasmuch as the
manufact ure is simplified . Wherever one may use slots with
a sma ll openi ng , a roun d core will be preferable, as it will
keep the additional losses to a minimum.


Gh. 20 Basic Machin e Designs

Table 20-1 (con ti nue d)

the st ator



Hornop ola r
Smo oth s t a tor core

c ,

'I'oo tl icd st ator COr e


Smooth st ator core

Uncapabl e of energy
conver sion in either

ene rgy conversion i n either d i r ect ion

ar wind ing' 1 can only be single -phase .

Some of the conce ivable machine designs are list ed in

Tab le 20-1. It gives combinations of heteropolar and
homopolar windings and toothed and smooth cores for the
st ator and ro tor. Combinations using ring windings and
claw-shaped cores are not included because such machines
are identical to those using a:single-phase heteropolar winding. For the same reason , there has been no need to include
cylindrical and toroidal heteropolar windings. The most
important of the modifications li sted in Table 20-'1 are
examined in the pages t hat follow.

Machines with One Winding

on the Stator and One Winding
on the Rotor

In a machine carrying one winding on the stator and one

winding on the rotor, electromechanical energy conversion
occurs mainly owing to variations in the relative position
of, and in the mu tual in duct ance between , the windings as
the rotor rotates . Variations in the self and mutua l inductance of the windings due to the saliency of the cores are of
secondary imp or t ance.
In the arrangement considered, only heteropolar windings
are used on the stator and rotor. The rotor core may be


Part Two. En ergy Conve rsion by Electrical Machines

t oot hed (notably , with sali ent pol es) a nd smoot h (rou nd
or cylindrical) . The stator core may likewise be toothed or
smooth. This gives a total of four combinations labelled as
A1, A2, B1, and B2 in .T able 20-1. Figure 20-1 shows a fourpole machine with single-phase heteropolar wi ndings on the
stator and rotor, and smooth stator and rotor cores (modification B2 in Table 20-1).
The winding curr ents i, or i 2 set up a four-pole (p = 2)
magnetic field (the figure onl y shows t.he magnetic lines of

Fig. 20-1 Machine wi th one stat or winding (1) and one roto r winding
(2) (PI



force due to i 2 ) . A plot of L 1 2 an d 11 (22) as functions of the

an gle I' between the ax es of the t wo windings is shown in t he
same figure . As is seen , the mutual inductance, proportion al
to the flux linkage of the magnetic fie ld due to i 2 with the
turns of winding I is a maximum when y = 0, that is, when
the axes of t he coil s run in the same direction . Wh en t he
axis of coil 2 m ake s wi th the axis of coil I an angle y = n /4,
which corresponds to a linear displacement along the periphery of the air gap through 1:/2 , or a quarter-cycle of change
in the field , t he Il ux linkage with coil 1 and t he mu tu al

Ch. 20 Basic Machine Designs


inductance will be zero . A cycle of change in the mutual

inductance is completed as the rotor moves through 2-r
or through a pole pitch an gle ,,?p = rr. In the general case , when
the windings set up a p-cycle field , the mutual inductance
undergoes a complete cycle of change as the rotor moves
through 21: or the pole pitch angle given by
,,?p = (2rrJ2:n;R) 2.. = 2:n;/p


If the rotor is ro ta ting a t angular velocity Q, the mutual

in ductance will alternate with a period given by

T = ,,?p/Q = 2n /pQ


Accordingly, the frequency of change in the mutual inductance, f, and the angular frequency of change in the
mu tual inductance, ro, are given by

1/ T = pQ/2:n;
w = 2:n;f = pQ


The sha pe of the plots for 12' 11' and 22 is typi cal of
round (cylindrical) cores with q = 1: a half-cycle of change
in 12 is triangular in shape, 11 and 22 are nearly constant;
i t is onl y wh en the slots in the stator and rotor are aligned
tha t the self-inductances show slight variat ions, bu t these
may safely be ignored . As the number of slots per pole per
phase, q, increases, the pattern of change in 12 takes on a
shape close to sinusoidal, which ha s a wholesome effect on
the performance of the machine . Thus, by increasing the
number of slots on a round rotor core carrying a singlephase winding the 12 pattern can be made nearly sinusoidal
in the round-core synchronou s machine shown as an example
of modifi cation A2 (see Table 20-1) .
A practically sinusoidal pattern of change in 12 can be
obtained with q = 1 as well , if the rotor cor e is so shaped
that the air gap at the too th axis is two-thirds to one-half
of the gap at its tips (or edges).
This ty pe of rotor (a sa li ent-pole rot or) with a si nglephase winding is used in modific ations A1 an d BL
Most frequ ently , electrical machines are built with singleor polyphase heteropolar windings (see Sec. 19-2) having
the same num ber of pole pairs. This is true of induction
ma chines an d conve nt iona l synchronous machines (see
Parts 4, 5 and 6 of this text) .


20- 3

Part Two. Energy Conversion by Electrical Machines

Machines with One Wind ing on the stator

and Toothed Rotor and Stator Cores
[Reluctance Machines)

In a machine with one winding on the stat or and toothed

rotor and stato r cores , electromechanica l energy conversion
is bas ed on the variations caused in t he self-inductance of
the winding by t he tee th on the cores. Figure 20-2 shows
such a machine with a single-phase concentra ted het erop olar





o I I Y I 7C/Z



Fig. 20-2 Machine with concentrated heteropolur st at or winding (1)

(PI = 2, ql = 1) and toothed stator (3) and rotor (4) cores with an
equa l number of teeth (Za = Z4 = 8)

windi ng on t he stator, and the rotor and stator cores having

the same number of t eet h, Zs = Z 4 = Z (modification A7 in
Table 20-'1). The tooth pitch angle of the rotor, '\'Z 4 = 2n/Z 4 , is
the same as that of the stator, '\'za = 2n/Za .
As is seen from the curves in Fig. 20-2, when Z:j = Z4'
variations in L l l are suffi ciently large for an effective energy
conversion t o t ake place. To avoid some undesirable effec ts
in operation, it will be well-ad vised to choose t he number of
teeth on the stator and rotor such tha t
Z 4 - Z~ = +2Pl
The rationale of such a choi ce will he expla ined la ter.
An example of a machine satisfying the condition in
Eq. (20-5) is shown in Fig. 20-3. The rotor has Z4 t eeth,

Gh. 20 Basic Mach ine Designs


where as the salient-pole stator .has 2Pl = 4 poles. The coils

of t he concentrated single-phase stator winding are wound
around the pole-pieces and laid out in major slots between
them. On the surface of the poles are made minor stator
teeth displaced from one~fianother by :a tooth angle I' Z3
= 2n/ Z;, where Z~ is the number of tooth angles that can

Fi g. 20-3 Machi ne with concentrated heteropolar stator winding (1)

(PI = 2, th. = 1) and toothed st ator (3) and roto r (4) cores with different numbers of teeth (Za = 12, Z; = 16, Z4 = 20)

be accommo dated rou n d the core . The number Z; must

satisfy the condition defined by Eq. (20-5) and be, of course,
a multiple of 2PI' that is Z~ = 2Pl (an in t eger) . This in
turn requires t hat the rotor should have a number of t eeth
which is a m ultiple of t he number of poles, t ha t is , Z4 =
2Pl (an integer). In our case,
Z4 = 2 X 2 X 5

= 20 - 2

X 2


= 16

For the machine t o oper ate normally, it is essent ia l that

each pole sh ould spa n 2/3 t o 3/4 of a pole pitch , t he rema inder being taken up by the major slots betwe en them. Accordingly , each pol e must carry an odd number, N 3' of mino r
te eth . This numb er must lie within the limits given above


two-thirds to three-fourths . of Z~/2p



Part Two. En ergy Conversion by Electrical Machine s

where Z~/2p is the number of minor tooth angles per pole

pitch. In our case,
N 3 = (2/3 to 314) X 16/4 = 3
Th e self-inductance of the stator wind in g, L Il = 'l\/i 1 ,
varies with the rel ative position of the stator and rotor
tee th. For the machine in Fig. 20-2 (with Z3 = Z4) ' it is
a m aximum when t he stator and rotor t eeth are ali gned,
for exam ple when '\' = 0, n /4, n /2, etc . It is a minimum
when a slot is oppo sit e a too t h , which happens when '\' =
= n18, 3n18, 5n18, etc. , and the permeance of the air gap
and the flux linkage 1Jf 1 for a given i 1 are minimal.
For the ma chi ne in Fig. 20-3, the self-in ductance of t he
winding is a maximum when the rotor tee th are align ed with
the stator tee th lying on the axes of the winding poles
(1, 1', 1", 1"'). This happens wheu v = 0, n/10, n/5, etc.
In this posi tio n , all minor teeth on t he pol es ar e approxim at ely opp osite the rot or te eth, and t he flux linkage of the
winding for a giv en conductor curre nt is a maximum. Conversely , t he self-inductance of the wind ing is a mi nimum
when the rotor slot s ar e aligned wi th t he st ator t eeth lyin g
on the axes of the winding poles, which happ ens when , say ,
Y = n/20, 3n/ 20, etc .
The self-inductance undergoes a complete cycle of chang e
as the ro tor moves t hrough one t ooth pitch , or one t ooth
angle Yz = Y Z 4' (The pol e pi t ch an gle of t he winding,
1'Pl = 2nlp l' and the tooth angle of the stator, Yz:i, ha ve
no effect on the perio d of change in the self-inductance .)
If the rotor is rotating at an angular velocity Q , the time
period of change in self-induct ance, it s frequ ency an d angular
fre quency are given by
T = Y'Z 4/Q = 2nlZ4Q
f = Z 4 Q/2n

= Z4Q


From comparison of Eqs. (20-8) and (20-3), it is seen th at

in syn chr onous reluctance machines, the frequ ency is Z 4/P
times that of conventional two-winding machines with th e
same rotational frequency .
Th e rationale of cho osing the number of t eeth subject t o
Eq . (20-5), may be explained as follows. For variat ions in
the self-inductance of the winding to be subs t antial, the


Ch. 20 Basic Mach in e Design s

rotor and stator teeth must take up the sam e relative position at every pole of the winding (this is true, for example,
of the case in Fig. 20-3 wher e the rotor and stator t eeth are
shown aligned at all the poles) . Let one of the rotor teeth
(say, too th No.1) be aligned with a stator too th at pole 1' .
Then the next adjacent rotor tooth will be displaced from
the nex t adjacen t stator tooth through an angle YZ3 - YZ 4;
the rotor tooth following it will be displaced from the corresponding stator tooth by an angle 2 (YZ3 - Yz,;) , etc. As a
result, the rotor tooth in alignment with the stator too th at
pole 1" will be separated from the first by an ang le YPl/2 or
YPl/2Y Z4 t ooth pitches of the rotor . Wi th the respect to the
nex t adj acent stator tooth , this rotor tooth will be displaced
through an angle YPI (Y Z3 - YZ 4)/2YZ4 whi ch must be
equa l t o t he t ooth angle of t he stator, that is
YPI (Y Z3 - YZ4)/2YZ 3 = +YZ3
Hence, on recalling tha t
1'111 =

2n lp 1

we obtain the cond ition defin ed in Eq . (20-5).

If stator winding 1 (Fig. 20-4) is a distributed one, and
the conduct ors carry ing cur ren ts in the same direction ar e
laid at each pole among several (q) slots (in Fig. 20-4, q = 3),
the stator core 3 need not be a t oothed one (modification B7
in Table 20-1). In the modification using a distributed win ding, variations in the self-in ductan ce of t he winding can be
produced by the teeth on the unwound core, 4. Th e slots in
core 3 carrying the winding may have a limited opening.
In the cir cumstances , core 3 may be treated as a smooth one.
Variations in the self-induct ance will be a maximum when
the rotor has the same number of t eet h , Z4' as there ar e
poles on winding 1 (in Fig. 20-4, 2p 1 = 4 and Z 4 = 4).
This design may be regarded as a specia l case of a machine
with the numb er of te eth meeting the condition defined in
Eq . (20-5) for a smooth st ator, when Z3 = 0, and Eq . (20-5)
redu ces to Z4 = 2p1'
In such a ma chine, the hi gh-perrneancc zones lie opp osite
the rotor teeth, and the low-permeance zones li e oppo site
the ro tor slots. If it has Z4 = 2Pl tee t h , such a rotor is called
salient-pole . The time period , frequency and angular frequency of var iations in self-ind uctance are give n by Eq . (20-8).
Apart from a heteropolar winding, the design in question
may use a single-phase homopolar winding (modification

Part Two. Energy Conversion by Electrical Machines


C7 in Table 20-1). In this design , variations in the self-indu cta nce of the win ding are obt ained by using a toothed core
for both the stator and rotor, the effect being a maximum
when Z3 = Z4' In performance, such a machine is not
unlike a single-phase heteropolar machine with Z3 = Z 4'




Ln o



Fig. 20-4 Machine with distribu ted heteropolar stato r (1) windi ng
= 2, ql = 3) and a toothed (salie nt-po le) ro tor (4) core (Z,I =




Th e t ime period, frequency and angular frequency of variations in the self-inductance of a homopolar winding are
found by Eq. (20-8).
The design examined in t his section is utilized in reluctance synchronous machines. It offers advantages of simple
construction and freedom from sliding contacts in the electric circuit of the winding.

M achines with Two Winding on the Stator

and Toothed Cores for the stator and Rotor
(Inductor M achines)

In t his case, electromechanica l '(or mechanoelectric) energ y

conversion occur s m ainly owin g to variations in the mutual
inductance between the s tator windings as a toothed rotor
core mov es relative to them .


Ch. 20 Basic Machine Designs

The win di ngs m ay be heter opolar and have t he same nu mber of pol e pairs (PI = P2) or a diff erent number of pole pairs,
or they ma y he homopolar . It is also possible to combine a
het eropolar an d a homopol ar winding. The stato r m ay be
built with either a t oothed or a smooth core. This leaves




Fig . 20-5 Machine with twu hc teropolur stat or windings, ] and 2

[12 = 2) and toothed cores for the s ta tor (3) an d rot or (4), wit h
th e same number of t eeth (Zs = Z4 = 8)

(PI =

us wi th five likely modificatio ns (A3 , B3 , C3, D3, and C5,

see Table 20-'1 ). One cycl e of ch ang e in the mutual in duct ance
is completed as the rotor (at 4 in Fig . 20-5) moves through
a t ooth pitch t Z 4 or a too th angle, '\'Z 4 = ,\,z.
If the ro to r is r otating at an angul ar veloci ty Q , the t ime
period and an gular frequency of the mu tual inductance can
be found by Eq. (20-8), assuming that the rotor core ha s a
number Z = Z 4 of te et h . According to the m ann er in whi ch
the mu tu al in du ct an ce between the win dings is m ade to
var y, the machine can be built in one of four modification s.
(i) The mutual inductance between the windi ngs varies
owing to changes in the mean perm eance of the air ga p wi th
the rot ation of t he to ot he d rotor core rel ati ve to the t oothed
s tat or core , bot.h having the sam e numb er of t eeth (modifica tions A3 an d C5 in T able 20-'1) . In this arra ngemen t , the
mutual induct ance has the same sig n in any po sition of the


Part Two. En erg y Conversion by Electrical Machines

rotor rel ative t o the stator . As the rot or rotate s, it oscillates

about its mean value , being a maximum when t he stator
teeth are aligned with the rotor tee th, an d a minimum when
the stator teeth ar e aligned with the ro tor slots (an d, of
course, when the s ta tor slots are aligned with the rotor
teeth) .
The numb er of t eeth

= Z3 = Z4


is chosen t o give the requisite fr equ ency of cha nge in the

mutual inductance, from Eq . (20-8) .
Similar va riations in t he mutual induct ance wi th Z3 = Z4
can be obtained with modification C5 which uses two homo pol ar wi ndings. It is equ iv alent in performance to modification A3 wh en the latter uses single-pha se windings with
PI = P2"
(ii) The mutual inductanc e between the two het eropolar
windings on the stat or is m ade t o va ry owi ng t o t he rotati on
of a t oot hed rotor core (modification B3 in Table 20-1). In
t his arrangement, t he sa liency of the sta tor is of minor signific an ce. For variations in the mutual inductance t o be as
large as possible, t he numb er of pol es Pi and P2 and the
numb er of teeth Z4 mu st be ch osen such t h a t


P2 + PI


For example, in the machine of Fig . 20-6 wi th a two-p ole

st ator winding 1 (PI = 1) an d a four-pol e stat or wi ndi ng 2
(P2 = 2), the condition defined by Eq. (20-10) will be satisP2 = 3).
fied if the ro tor h as t hree teeth (Z 4 = PI
In su ch a machine , the permeance will be a maximum in
zones B I , B 2 , and B 3 whi ch are ali gned wi th the ro to r teet h.
It is an easy m atter to prove that t he mutual induct an ce
between the stator windi ngs is a function of the posit ion
that zones B I , B 2 , and B 3 take up relati ve to thes e windings.
When the ro tor takes up t he posi tion sh own in t he figure
(the angle i' between winding 1 and the rotor te eth is n /G) ,
the mu tual induc ta nce 1 2 is a posi ti ve maximum (the magnetic fie ld du e to the cur rent i 2 pro du ces a maximum positi ve
flu x linkage with winding 1). If we rotate the rotor through
n /3 , its tee t h will move in to the position previou sl y occupied by the slots an d, as can read ily be sh own , the flux
linkage and the mutual inductanc e will change sign and
take ea ch a m ax imum negative va lue equal to the positive

Ch. 20 Basi c Machine Design s


maximum va lu e in magnitude. With PI = 1 and P2 = 2,

the av erage mutual inductance will be zero. It will likewise
be zero if P2/Pl is an even number, that is , if t he windings
are such tha t given smooth cores, the mu tual inductance


Fig . 20-G Mach ine with t wo hotoropolar sta tor windings , 1 and 2,
with a diff eren t num bel' of pol e pairs (PI = '1 , P 2 = 2), a smoot h
stator core (3), and a toothed rotor core (4). Z4 = PI
P2 = 3

between them is zero . If P2/PI is an odd number, the average

mutual inductance will be non-zero, and the mu tual inductance will be pulsating about its mean value .
The number of rotor teeth Z 4 is uniquely fixed by the
specified frequency and angular frequenc y . As a rule Z 4
is fairly large , and in order to satisfy Eq . (20-10) , winding 1
must be made with a moderate numb er .of pole pairs (PI
= 1, 2, 3) and winding 2 with a large number of pole pairs,
clos e to that of rotor te et h , P2 = Z 4 - Pi - This introduces
some difficulties in t he m anufacture of winding 2. In fact ,
if Z 4 is very large, one has to use the modification described
in (iii) below . Winding 2 is laid in the minor slots shaped so
that t heir effect on the permea nce of t he air-gap may be
neglect ed . Winding 1 is laid in the major slots whi ch can be
formed by enlarging the cross-section of som e minor slots
without increasing t heir total numb er (Fig. 20-6) , or t hey


i)urt Two. Energy Conversion

by Electrical Machin es

may rep lace a group of several minor slots and te eth by

removing one or several coils from winding 2. When Z 4 =
= 2PI' the two windings have the same number of pol e
P2 = Z 4 - PI = 2PI - PI = PI
If, however , this arrangement is to perm it variations in the
mutual inductance , one of the windings mus t be distributed
or, if both windings are left concent rat ed, they must be
displaced from each other by a qua rter of a cycle, as shown
in .t he cross-sectional view of modification B3.
(iii) The mutual inductance between the two heteropolar
windings on the stator is made to vary owing to changes in
the position of a toothed stator core relative to a toothed
rotor core, both having the same number of teeth . In this
arrangement, the number of te eth on the stator and rot or
must be chosen such tha t there are as many high-permeance
zones as in a machine with a smoo th stator core in (ii )
(Fig . 20-6), that is, P2 + PI (when P = 1 and P2 = 2, the
number of such zones will be 2 + 1 = 3 or 1).
To obt ain this number, it is essential that the difference
in the number of te eth between the stator and rotor be
P2 + PI'
Z 4 - Z3 = P2 + PI
To prove, at the centre of a high-permeance zone , say , B I
in Fig . 20-7, a rotor tooth is opposite a stator tooth (or, whi ch
is t he same , a rotor slot is opposite a stator slot) . An adj acent rotor tooth is displaced from an ad jacent stator too th
by an angle I' Z3 - I' Z4; the nex t adjacent rotor tooth is
displaced from the next ad jacent stator to oth by an ang le
2 ('\'z3 - YZ4) ' and so on . To arrive at the centre of the
next high-permeance, say, B 2 the displacement must be
2n /(p2 + PI) or Z 4/(P2 + PI) roto r teeth. Then , because at
the centre of zone B 2 the rotor too th mus t again be opposite
the stator tooth, the displacement of this rotor tooth from
the correspond ing stator tooth
Z 4 (YZ3 - YZ4) /(P2 + PI)

mu st be equa l to the tooth angle of the stator, that is,

Z4 (-\'Z 3 - '\'Z 4) /(PZ + PI)


Hence, the number of tee th on the stator and rotor must

sati sfy the condition defin ed by Eq . (20-11).

., .


' .



Gil. 20 Basic Machin e Designs

Also, for windings 1 and 2 to be able t o form balanced

circuits, Z:i must be equal to 2p z (an integer) , if
PiAs an example, Fig. 20-7 shows a machine having p z = 2
and PI = 1, as does the machine in Fig. 20-6. In view of
the desired frequency ,


Z3 = 2 X 2 X 5 = 20
Z 4 = 20 + 2 + 1 = 23
The pattern of changes in L 12 with t he angular posi tion
of the rotor is qualitatively the sam e as for the machine in






Fig. 20-7 Machine wit h two heterop olar stator wind ings, 1 and 2,
differing in the number of pole pair s (PI = 1, p z = 2). The stata l'
and rot or are built with t oothe d cores differi ng in the num ber of teeth :
Z4 = Zs

+ PI + pz =


+ 1+2=


Fig. 20-6. But the self-inductance varies nearly sinusoidally ,

and it completes a cycle of change in one tooth angle , I'Z4'
It is to be noted that as the rotor moves through one tooth
pitch or through I' Z4' the axis of a high-permeance zone moves
through an angle 2'Jt/(Pz + PI) , so that zone B I takes up the
place of B z , zone B z takes up the position of B 3 , and so on.
As is seen , the high-permeance zones rotate at a higher spe ed


Part Two. En ergy Conversion by Electric al Machin es

tha n the rotor by a factor of

2n/Yz 4 (p z + PI)

Z4/(PZ + PI)

Th e t ime period and fre que ncy of vari ations in the selfinduct ance can be found from Eq . (20-8).
(iv) The mu t ual induct ance bet ween a heteropolar an d
a homopolar winding is m ade to var y by the rotation of

F ig. 20-8 Machine with a},heteropolar (2) and a homop olar (1) st at or
windings, with a smoot h stator core](3) l and a 'ltoothe d rotor core (4).
Z,j = PZ = 2

a to othed rotor core relative to a smo ot h st ato r core (modification D3 in Table 20-1) . As is seen from Fig. 20-8, the
het eropolar winding is lai d out in t he stator slo t s (in the
figur e, p z = 2). The homopolar .winding is wound as a ring
around the rotor shaft. The figure shows the posi tive directi ons of the currents in t he windings and the magnetic fi eld
set up by i 2
If the stat or and ro tor cores were smooth, t he magne tic
field established by win din g 2 would be periodic (as shown
in Fig. 20-1). Its lines would close vi a the yokes and li nk
the cur rent i 2 in the slots wi thout linking with winding 1 . To
make variati ons in L 12 as large as pr acticable, the rotor core

Ch, 20 Basic Machine Designs


is made with teeth, and the number of teeth is taken equal to

the number of pole pairs for winding 2, that is,
Z 4 = P2
As an example, the condition defined by Eq. (20-12) will be
satisfied by the machine shown in Fig. 20-8 (P2 = 2) when
Z4 = 2. When the rotor is in the position shown in the figure
(the angle between the axis of the 8 pole on winding 2
and the axis of a rotor too th is 'Y = 'Yzj2 = n/2), the mutual
induct ance L 12 betw een windings 1 and 2 has a ma ximum
negative value. In this position, the high-permeance zones
aligned with the rotor teeth lie opposite the N poles on winding 2. Conversely, the zones lying opposite the 8 poles have
the lowest permeance, so the periodic magnetic field whose
lines close around the cur rents in the slo ts, along the yokes,
and across t he gaps is insignificant.
In contrast, th e homopolar field whose lines close around
the coil ends on the N poles of winding 2, across the gap
zones having a maximum permeance, across the yokes,
through the shaft, end-shields, an d frame (see Fig. 19-1d) is
substantial (it is shown by dashed lines in Fig. 20-8). As is
seen , the lines of t his homopolar fiel d link with homop olar
winding 1, and the resultant flux linkage is neg ative. If
we turn the rotor through 'YZ4/2= n/2, the axes of the rotor
teeth will line up with the axes of the 8 2 poles on winding 2,
the high-permeance zones will lie opposite the 8 2 poles on
winding 2, and the result an t homopolar fiel d will produce
a positive flux linkage with winding 1. It should be noted
that in Fig. 20-8 we have chosen small values for Z4 and P2
only to simplify the illustration. In practical machines, t he
relationship between f and Q is usually such that Z4 = P2
must be fairly high. As already noted , winding 2 with a large
number of poles is difficult to make. In fact , if Z4 = P2
turns out to be too large, the modification being discussed'
has to be replaced by that examined in(v)below.
(v) The mutual inductance between a heteropolar and'
a homopolar winding is made to vary by the rot ation of
a too thed rotor relative t o a toothed st ator, having differen t
numbers of tee th (modification C3 in Table 20-1).
As follows from Fig . 20-9, this modification differs from
that in (iv) only in that the stator core has teeth. To make
variations in L 12 as large as practicable , the number of teeth
on the stator (Z3) and the rotor (Z4) must be chosen such"
15-0 16 9


Part Two. En erg y Conversion by Electric al Machines

that the number of high-permeance zones form ed around the

periphery of the air gap be equa l to Pz, as in the m achine
shown in Fig. 20-8. As has been pro ved in (iii) , the numb er
of high-perme ance zones for a t oothe d stator and a toothed


Fi g. 20-9 Machin e with a heteropolar (2) and 'a h omopol ar (1) wind ,
in g on th e stator an d toothed st at or (3) an d rotor (4) cores h aving a
different number of t eeth (Z4 = Z 3
P2 = 12
2 = 14)

rotor is equa l to the diff erence in the number of teeth between the stator and rotor. Therefore , Za and Z4 must be
chosen such that
As an example, for the machine in Fig. 20-9, which uses
winding 2 with two pairs of poles (Pz = 2) and st ator 3 with
Za = 12 t eeth, the condit ion defined in Eq. (20-13) will be
sa t isfied when

Z4 = Za + pz = 12 + 2 = 14
When the roto r takes up the position show n in the figure
(the angl e between t he axes of the st ator and rotor t eeth is
I' = 31'Z4/2 = 3:rt/14), the mutual inductance between
windings 1 and 2 has a maximum negative value. In this position, the high-permeance zones in the air gap, wher e the

Ch. 21 Unidir ection al Energy Conversion


stator teeth lie opposit e the rot or teeth , are aligned with
the N2 pol es of winding 2. In contrast, t he zones lying opposite the S poles have the lowest permeance. If we turn the
rotor through an angle "Yz 4/2 = :rt/14, the axes of the ro tor
teeth will ali gn themselve s with those of t he stator tee t h on
the axes of the S poles. The high-p erm ean ce zones will then
lie opposite the S2 poles on winding 2 and produce a homopolar magnetic field which links wi th winding 1.
The design with two windings on the stator is frequently
used in special-purpose machines. Among its advantages are
the relatively high frequency of variations in the self or
mutual inductances at a relatively low rotational speed,
and also freedom from sli ding cont act s in the elect ri c circuits
of the windings (for which reason such machines are called
brushless or contactless).
In t he generator mode of operation, such m achines generate
voltages at a high frequency, although the rotor is rotating
at a medium velocity (in ductor generators) . In the motor
mode of operation, their rotors rotate at a substantially lower speed than the m achines having windings on both the
stator and rotor. Because in such motors the ro tor speed is
reduced electromagnetically (without any gearing), they
may be called electromagnetically down-geared motors.


Conditions for Unidirectional Energy

Conversion by Electrical Machines


The Single-Winding M achine

In this chapter, we shall discuss what cur rents the windings

of a machine must carry for un directional energy conversion
to take place. The discussion will be concerned with the
same machine designs as are listed in Sec. 20-1.
To begin with, we sh all turn to the equat ion of electromagnetic torque, Eq. (18-6), for a single- or a two-winding
machine. In this equation, the self-inductance of one winding, L 11l or the mu tual inductance L 12 between two windings
is a periodic function of t he angular position of the rotor
or time. For unidirectional energy conversion, the currents
in the windings must vary so that the mean electromagnetic
torque is nonzero.
15 *


Part Two. Energy Conversion by Electrical Machine s

Because all the events involved recur periodically, it

will suffice to determine the torque averaged over a period




and to define the conditions

for currents under which elect-r~r---/----II--....Y....---..;..t ric energy is converted to
mechanical (To> 0) or back
(To < 0).
Let us consider a single-winding machine first (see Sec. 20-3).
As follows from Eq. (18-6),
when n = 1 and k = 1, the
electromagnetic torque developed
by a machine with
one winding 1 carrying a curI
rent i 1 (see Figs, 20-2 through
20-4) is given by

1 ' 2 dL ll
T --2~1~


No matter where the winding is wound (on the stator or

rotor) and how it is arranged, variations in L 11 will be
qualitatively the same, with
---7-an angular period (tooth angle)
To (generator)
'Y z or a time period T
= -2n/pzQ = 2n/ro [see Eq.
\ ,I, rem (geflerator)
(20-8)], about some mean selfFig. 21-1 Conditions for unidi- inductance, L 11 o Expanding
into a Fourier series and retainrectional energy conversion in
a single-winding machine
ing the zeroth and first terms,
variations in the self-inductance with time may be described by an equation of the
L 11m cos rot
L 11 ~ L 110
- ,/

where ro = Qpz = 2n/T is the angular frequency of variations in the self-inductance (see Fig. 21-1).

Ch. 21 Unidi rection al Energy Conversion


The derivative of the self-inductance with respect to the

an gular posi tion of the rotor is
dL ll/dl' = (dLll/ dt) (dt/ dl') = - PZL ll m sin wt
(21 -4)
where I' = Qt , an d dt/dl' = '1/Q.
As we have already learned , t he current in the onl y windin g of a m achine must be an alternating one . Using Eq. (21-1),
it is an easy matter t o see t hat if the wind ing carri ed a const an t cur rent, the mean t orque T o would be zero.
Let us li mi t ourselves t o the fund ament al component of
current, responsible for the la rgest me an electromagnetic
torque. Then,
i i ~ 1 1m cos (WIt

N ow the qu estion is: What should WI and cp be for Toto

be a maximum, with all other con dit ions being equal?
Since the mean torque is given by

To =

PZI~:;/l1m ~ cos- (Wit + rp) sin wt dt

its ev aluation reduces to evaluating its in tegral. Upon trigonometric manipulations in t he integr and , we get

~ cos'' (Wit + (p) sin wt dt


= ~ ! .\

si n tot dt

+-} ~ cos (2Wit + 2cp) sin wt dt

The first t erm on t he right-h an d si de is equal to zero.
The second term may be re-written as

4- Jsin [(W+ 2W t+2 cp] dt



.\ sin [(co - 2Wi) t - 2cp] dt

When the angul ar frequency of the current is
WI = w/2
the time period of the current, T I = 2T , is twice the time
period of the self-inductance , and the period of the current


Part Two. Ene rgy Conversion by Electrical Machines

squ ared , T l /2 , is equal to the time period of the self-induct ance (see Fig. 21-1). Then t he mean electromagnet ic torque is
To = (p zI\mLnm/8) sin 2cp
When (p = n /4, the mean torque is a maximum in the
generator mod e of operation . When (p = - n/4, it is a ma ximum in the motor mode of operation . Also, the periodic
component of the current squared is in phase with the selfinductance in the former case , and in anti-phase in the latter
Th e respective plo ts of cur rents and t orques appe ar in
Fig. 21-1. Th e angle cp = n /4 corresponds t o a time lead of
t = CP/COi = T l /8.
To sum up , it may be argued th at for unidirectional energy
conversion, a sin gle-windi ng machine must carry a current
at angular frequency COl equa l to h alf th e angular frequency
of variations in the self-inductance:
CO l

= co/2 = P zQ/2

The direction of energy conversion dep ends on the phase

ang le between the cur ren t and the self-inductance. When
cp = sil, the machine will be operating as a motor. When
cp = - n/4, it will be operating as a generator.
Th e ang ula r veloci ty of the ma chi ne is proportional to the
angular frequency of the cur rent in t he win ding connected
t o the electrical sys tem
Q = 2COl/P z
A machine whose angular velocity is proportion al to the
angular frequency of the electrical system will be called
a sunchronous machine . A ma chine whos e angular velocity
does no t satisfy this rela t ion will be call ed an asynchronous one .
From Eq . (21-8) it follo ws then that all single-winding
a .c. machines are synchronous machines.

Two-Winding M achines

Th e electromag net ic t orque developed by a two-winding

machine is giv en by

T = i 1i z dL i2/d 1,

+ ~. i~ dLii/dl' + ~ i ~ dLz)dl'

irrespective of the winding arrangement.

(21 -9)

Ch. 21 Unidire ctional En ergy Conversion


A major contribution to the electromagnetic torque comes

from var iat ions in the mutual inductance between the
windings and is represented by the first term in Eq . (21-9)* .
Therefore, we may li mi t our ana lysis t o the first term.
As in the pr evious case , we may li mi t ourselves to the
peri odic component of the self-inductance with an angular
peri od YI> (or Yz) and a time period
T = 2rr,/PoQ = 2rr,/w
where P o

P is the number of pole pairs on t he heteropolar

windings of the stator and rotor [see Eqs . (20-2)
t hrough (20-4)]
p z = Z is t he number of teeth per pole on the rotor
of a machine with two stator windings [see
Eq . (20-8)]
As in the previous case, the m utual in duct ance varies
with time as
L 12 = L 12 0
L 12 m cos tot

wher e ro = Qpo = 2rr,/T.

The deriva ti ve of the mutual inductance with respect to
the angula r position of the rotor is

dL1Z/ dl' = (dL 12 /dt) (dt/dy) = - PoL 12 m sin wt


In t he gen era l case, the windings carry alternating currenls** . Limiting ourselves to the fundamenta l components
as contri buting most to the electr om agnet ic torque, we may
t, = J Im COS (WIt
12 = J 2m COS (W2t


In Fig. 21-2, WI = 4w, W 2 = 3w, cri = 0, and (P 2 = - rr,/2 .

Now le t us find the va lues of WI' w 2 , (PI' andtp, that will
lead to a maximum mean electromagnetic' torque in
Eq . (21-1), with all oth er conditions being equal.

* Th e other components of the mea n torque can be found as for a

single-winding ma chine. As follows from Eq . (21-6), th e mean torque
of t his kind m ay be non-z ero at WI = w/2 or W 2 = w/2. Th en i t will
be a maximum a t (jlI = rtl4 or rp 2 = 'It/4.
* * A m achin e with, say , the second winding carrying d .c, is a specia l case for whi ch W 2 = 0 and i 2 = canst .


Part Two. En ergy Conversion by Electrical Machines

The mean torque is given by


To=-(Po[lm[2mLI2mIT).\ cos ((Oit+CJli) cos ((02t+ (P2) sin wt dt

The products of the cosines in t he integrand may be rewritten

"2 cos [((01 + (02) t + (PI + CJl2] + 2" cos [(Wi - (02) t + CJlI - CJl2]

where the first term varies with a frequency W I

00 2 , and
the second term with a frequency WI - (02'
If one of these frequencies is the same as the frequency of
variations in the mutual inductance, that is,

+ 002 =



WI -


= (0


then the mean torque will be nonzero. To demonstrate, on

replacing the products

~ cos [(WI + (02) t +

(Pi + CJl2] sin wt


2" cos [(001- (02) t + CJli- CJl2] sin wt

by a sum of four trigonometric functions, we get

To = (Po[imI2mLi2 m/4T) .\ sin r(Oi + (02- (0) t + (P I + CJl2] dt

Hence, on satisfying the condition defin ed by Eq. (21-13),
we obt ain
where the " +" sign applies when WI + (02= 00, and the "-"
sign applies when WI - (02 = (0.
The integrals of the remaining three terms of the sum,
varying at frequencies
(01 + (02 + (0 =1= 0
WI -


are equal to zero.


+ 00 =1= 0

002 -


=1= 0


Ch. 21 Unidirectiona l Energy Conversion

As is seen from Eq. (2'1-'14), in the motor mode of operation (To > 0), the mean to rque is a maximum when qJl
+ qJ2 = n/2 ; in t he generato r mo de of oper ati on (T o < 0),
this happens when qJl + qJ2 = - n/2 .
Unidirectional energy conversion by a t wo-winding machine is illustrated in Fig. 2'1-2. With the frequencies and
ph ase adopted in the figure,
qJl -

CjJ 2

= n /2

so the resultant torque is nonzero .

To sum up, for unidirectional energy conv ersion by a twowinding a.c. machine, it is
essen tial that t he sum or the
difference of the angular frequencies of the ' currents in the
windings be equal to the angular fr equency of variations in
t he mutual inductance between the windings. The direction
of energy conversion is determined by the magnitude of the
sum or difference of the phase
angles of cur rent s with respect
t o the mutual inductance.
Whe n 0 < ((PI + qJ2) < n/2 ,
electric energy is convert ed to
Fig. 21-2 Condi tio ns for unid imechanical;
when - n/2 <
rectional energy conversion in
(rpt + qJ2) < 0,
a two-w inding machi ne
ene rgy is converted to electric.
Accor din g as t he ro t at ion al frequency of t he rotor does or
does not change with variations in the ext ernal torque, there
may be asynchronous machines and synchronous machines.
In a synchronous ma chine, both wind ings carry currents
whose ang ul ar fr equ encies are fixed in ad vance. In the
gener al case, t he m achine conv erts th e electri c energy fed into
two windings. Therefore , such u n i Ls are also call ed doublejed machines. With COl and CO2 held constan t, the angular
velocity of t he rotor in a synchronous machine rema ins con -

Part Two. En ergy Conversion by Ele ctrical Machin es


stant, irrespective of t he t orque on its shaft *





= constan t

Most frequently , synchro nous machines are built with

a three-phase heteropolar win ding on t he sta to r and a singleph ase heteropolar winding on the rotor. If we pu t W 2 = 0
and CjJ 2 = 0, all t he relati ons derived above will full y apply
to such a ma chine.
If not otherwise qualified, the te rm "syn chronous" refers
exactly to the above type of machine.
In an asynchronous machine (primarily, a motor) only one
winding, sa y 1, is connected t o a lin e whose frequency , say WI
is fixed in advance . The other winding is either shortcircuited or connected acro ss an impedance , and the cur rent
i 2 in this winding is produced by electromagnetic induction.
Accordingly, asynchronous ma chines are more frequently
called induction machines . . The mutual emf is given by
e12= -d'P'12/ dt = - { i 1m L 12m (W1-W) sin [(W1- W)


i1mL12m (W1 + w) sin [(W1


'1'1 2 = i I L 12 = i l m cos

( W It

t+ {pd

+ o) t + {Ptl
+ (PI) L I 2m cos to t

is the mu tual flu x li nk age.

The frequ ency W2 = Wi - W of i 2 is a fu nction of the
angular velocity of the rot or (W2 = WI - Qp o) and satisfies
the condition for unidirectio nal energy conversio n defined
in Eq . (21-13).
Most frequently , asynchronous (induction) m achines are
built wi th a three-phase het eropolar a .c. winding on the stator, and a three-ph ase (or poly-phase) heteropolar shortcircuited winding on the. rotor .
If not otherwise qualified , t he term "induction ma chine"
refers to the above type of m achine.

* Vari ation s in th e load on th e shaft brin g about only cha nges in

the amplitu de and phase of it and i 2


Ch. 22 Windings for A.C. Machines

22 .

W indings for A.C. M achines

22- 1

introductory Notes

Our discussion will be limited t o heteropolar cyli ndr ical

windings since they ar e used most frequently in electrical
The coils of such windings are usually laid out in slots on
the stator or rot or.
The arrangement of single-layer (Fig. 22-1a) and tw o-layer
(Fig . 22-1b) he teropolar windings along with that of a sim ple polyphase single-layer wi nding has been explained in
Sec . 19-2. Therefore, our discussion here will only be concerned with pol yphase two-layer windings with m 1 phases,
since they are used most freq uently in a.c. m achines .

The Structure of a Polyphase

Two-layer W inding


A two-layer m-phase wind ing is designed for connection to

an m-phase ba lanced a .c. li ne or system . In the case of
a three-phase system, they can be connected in a star or
. a delta (Fig . 22-2a and b).
For the phase currents l A' I B, and I c to form a ba lanced
set, it is essentia l that the ph ase windings should have t he
same inductive reactances . Thi s req uirement will be satisfied if the axes of the phase windings are displaced from one
ano ther through an ang le equa l to 11m of the angular period
of t he field (the pole pitch ang le)*


The core of a three-phase , two- la yer winding (m = 3) is

shown in Fig. 22-3. Each ph ase winding consists of several
coils (Fig . 22-4a), one coil side lying in the top half of
a slot, and the other in the bottom half of another
slot about one pole pitch away . Each coil may have one
turn (we = 1) or several turns (we > 1) insulated from

* Thi s equa tion applies whon

lire displaced through Yp/4 .



2. Wh en


2, th e ph ases

I~ '


Part Two. Energy Conversion by Electrical Machin es

Fig. 22-1 Windings: (a) single-layer and (b) doubl e-layer



Fig. 22-2 Three-pha se winding: (al star-connected and (b) delt aconnecte d





Part Two. Energy Conversion by Elect rical Machin es

one another and from the slot sides (the coil in Fig. 22-4a
has two turns). The coils may be lap-wound or wave-wound.
In a lap winding, each coil is connected to the next adjacent
coil in series. In a wave winding, each coil is conn ected to
a coil two pol e pitches farther away than the nex t adjacent
Each coil has two leads. Let the lead on the upper coil
sid e be t he st art (8) of the coil. Th en the lead on its lower
side will be its fin ish (F). As a rule, the lead s of a coil are


.-...,, ~

-S 1




Fig. 22-4 Coils of a double-layer lap winding (solid lines) and a double-layer wave winding (dashed lines): (a) actual arrangement ; (b)

made long enough for direct connection to another coil.

According to the manner in which the coils are interconnected
within a phase, the leads may be differently shaped and proportioned.
In diagrams, coils are usually shown as single-turn loops.
As is seen in Fig. 22-4b , a wave-wound coil differs from
a lap-wound coil only in that the leads are sh aped differently (the leads of a wave winding ar e shown by the dashed
lines). The coil pitch y equal to the distance between the
coil sides, may be equal to a pole pitch (y = r), or it
may be somewhat shorter than one pole pitch, or chorded
(usually, y = 0.8'];). Accordingly, there may be a full-pitched
or a short-pitched (or chorded) winding.
The pole pitch and the coil pitch may be measured in terms
of the distance along t he periphery of t he air gap or in

Ch. 22 Windings for




tooth pitches

= Z/2p (tooth pitches)

y = yc/tz (tooth pitches)


where Yc


= coil (or slot) span (see Fig. 22-4)

= 2'JtR/Z = tooth pitch

Z = number of teeth (slots) on the core

Taking as an example a three-phase, four-pole, shortpitched (chorded) winding with y = 7 and Z = 36, Fig . 22-3
shows how the coils should be distributed among the phases,
the coil sides laid out in slots, and the positive currents
directed in the coil conductors of polyphase, two-layer
The total number of coils in the winding is equal to the
number of slots . So each phase contains

Z/m = 36/3 =



To establish a four-pole field, the coils in each phase

should be divided into 2p = 4 groups uniformly distributed
all the way around the circumference (one group per pole
pitch). Each group contains q = Z/2pm adjacent coils.
The number q is equal to the number of slots per pole per
q = Z/2pm = 36 -;- (2 X 2 X 3) = 3
Let us designate coils by the Nos. of the slots in which their
top sides are laid. Then phase A will include the following
coil groups: (1, 2, 3), (10, 11, 12), (19, 20, 21), and (28,
29, 30).

Adjacent groups in a phase are displaced from one another

by one pole pitch

= Z/2p = 36/4 = 9 slots

For the resultant field to be periodically varying, all the

coils in each phase must carry identical currents reversing
in direction as they pass from one pole pitch to the next.
Assuming that the current in phase A (see Fig. 22-2) is in
the positive direction, the currents in the top sides of coil
group (1, 2, 3) will he flowing "inwards" (away from the
reader), the currents in the top conductors of coil group
(10, 11, 12) will be flowing "outwards" (towards the reader),
etc. To facilitate design work, the coil groups in which the


Pa rt Two. Energy Co~version by Electrical Machines

top conductors carry curr en ts flowing away from the reader

are assigned the index of the start of a given phase, A, and
the coil groups in which t he top conductors carry curren ts
flowing towards t he read er are ass igned t he index of the
finish of the sam e phase, X.
Given the same positive directions of cur rents, the patterns
of coils and currents in the remaining ph ases will be t he
same as in ph ase A. Th e only difference will be that phase
B will be displaced from phase A counter-clockwise by
an angle
Yplm = 2nlpm = nl3
that is, through 2'r:!m = 18/3 = 6 slots. In turn , phase
C will be displaced through the same angle from phase B.
If the coils of a phase are divided into a iden tical parallel
paths (circuits) within each of which they are connected in
series, then each parallel pa th will carry a current
equal to IIa.
Referring to Fig. 22-3, it is seen that the currents in both
the top and bottom con ductors of a phase set up patterns
repeated every four poles, th at is with a period p = 2, so
that in a short-pitched winding the currents in the bottom
layer are replicas of the currents in the top layer, displaced
by 't - Y = 9 - 7 = 2 slo ts clockwise. If the winding
were full-pitched (y = 't), the layers would not be displaced
from each other, the currents in the top and bottom conductors
in all the 'slots would be in t he same direction and the conductors of a given phase would take up q slots per phase.
In short-pitched (chorded) windings (see Fig. 22-3), the
phase conductors are laid out in q
(r - y) = 3
- 7 = 5 slots per pole. Chording results in an expanded belt
occupied by the phase conductors within each pole pitch
and, as will be explained in Sec. 24-5, makes the air gap
field more sinusoidal.


Connection of Coils in a Lap Winding.

The Number of Paths and Turns
per Phase

As already noted, the coils of a winding may be lap-wound

or wave-wound. In a lap winding, each of the q coils within
a given pole pitch is connected to the next adjacent coil ill

Ch. 22 Windings for A.C. Machines


series ai ding to form a coil group. For example, connecting

the finish of coil 1 to the start of coil 2, and the finish of
coil 2 to the start of coil 3 (Fig. 22-5) produces a phase A

- .:::::::::: c


~ ....--.-




Fig. 22-5 Coil connection in a lap win di ng (Z
q = 3 , 1" = 9, !J = 7, a = 1)

36, p

2, m


coil group consis ting of coils 1, 2 and 3. The other coil

groups in phase A, (10, 11, 12), (19,20,21), and (28, 29, 30),
are formed in a similar manner.
The start of a coil group is the start of the lowest- numbered coil, and the finish of this coil group is t he finish of the
highest-numbered coil. For example, the starts of the coil
groups listed just above are the starts of coils 1, 10, 19 ,
and 28, whereas the finishes of the coil groups are the finishes of coils 3, 12, 21, and 30.
16- 0169


Part Two. Energy Conversion by Electrical Machines

The term "lap" refers to the fact that, in going around

from the start of a coil group towards its finish, the previous
coil overlaps , as it were, the next adjacent one (see Fig. 22-5).
The leads of the coils in a la p winding provi ded for connection
to the next adjacent coils are; shown in Fig. 22-4. In the
simplest case, when t here is only one path (or circuit) per
phase (a = 1), the coil groups in a lap winding are connected in series. This arrangement is shown in Fig. 22-5.
For this winding, the numbers of coils, phases and pole
pairs have been chosen the same as for the winding in
Fig. 22-3. For better presentation , the top coil sides are
shown displaced counter-clockwise from t he bottom coil sides
l aid' in the same slots . (The coil No . is the same as that of
the slot where t he top sid e is lai d.)
For proper periodic variations in the currents carried by
the coil sides of the ph ase, coil gro up (1, 2, 3), bearing the
index A, is conn ected in seri es opposition with coil group
(10 ,11,12), bearing the index X . The finish of group A is
connected to the finish of group X . The start of group
(10, 11, 12), bearing the index X , is connected to the start
of group (19, 20, 21) bearing t he index A, and so on .
If the positive directions of currents in the coils are chosen
in advance (see Fig. 22-3), it is an easy ma tter to establish
the sequence of connection for the coils .
In the other phases, the coils are interconnected in the
same manner as in phase A .
The coil groups in the lap winding of Fig. 22-5 can be
interconnected in a simpler way . Referring t o Fig. 22-6,
the coil groups are shown as sectors spanning an angle
yp l2m subtended by the top coil sides. The numbering of the
coil groups is given within the sectors. Each group has two
leads. The start of a coil group is the lead of the lowestnumbered coil in t he group.
When a . 1, the coil grou ps in Fig . 22-6a are connected
in the same manner as in Fig . 22-5. The arrangement in
Fig. 22-6b differs in t ha t the same coil groups within each
phase form the largest possible numb er of paths, a = 2p = 4.
Each path in a phase is formed onl y by one coil gr oup . The
positive directions of cur rent in the coil groups are indicated by arr ows at the cur ren ts I Ala , I !. and I cia (pointi ng
away from the finish towards the start of the forward groups
A, Band C, and from the start towards the finish ill
the backward groups X, Y,- and Z ).

Ch. 22 Windings for A.C. Machines


~ IT o

obtain the desired positive directions for the coil currents the starts of groups A, B, and C must be connected to
the starts of the respective phases (A, B, and C), their finishes to the finishes of phases X, Y, and Z, respectively.

Fig. 22-6 Coil group connection in the lap winding of Fig. 22-3 for
various numbers of paths: (a) for a = 1, (b) for a = 2p = 4

Conversely, the starts of groups X, Y and Z must be connected to the finishes of the respective phases, X, Y and Z,
whereas the finishes of these groups to the starts of phases
A, Band C.
There is also a way of connecting the coils in a winding
where a ranges between 2p and unity. Now, the coils are
connected in series-parallel. For example, in the arrangement of Fig. 22-3, with a = 2, each path will contain two
coil groups. Generally, the number of coil groups per path
is 2p/a. This number must always be an integer. For the
current to be equally shared among the paths, the latter
must be completely identical (that is, present the same
resistance and inductive reactance). This requirement is
satisfied, if the paths are assembled from the same number
of properly interconnected coil groups and have the same
number of series-connected turns
W = (2p/a) qWe
where We = number of turns per coil
qWe = number of turns per coil group
2p/a = number of series-connected coil groups per path.
16 *



Part Two. En ergy Conversion by Electrical Machines

Coil Connection in the W ave W inding

The distribution of coils among the ph ases and t he choice

of positive directions for the currents in the coils ar e independent of whether the coils are connected in a lap or in
a wave winding. Therefore, with given Z , p, y , and In,
the coil structure (say, the one shown in Fig. 22-3) is equally

~7-f--~ ~~A


f /

/ 41 / 25 ;If / \
--...,..:::><~:-f /2 4 -1/ \
" jt "23 / \
" ,;:


Fig. 22-7 Coil connection in a wave winding (Z' = 36, p

3, q = 3, 1: = 9, y = 7)



applicable to both a lap and a wave winding. In a wave

win ding, however, t he phase coils are in terconnected differently. Here, the winding progresses around the cor e by
passing successively under each pole before again approaching the st ar t ing point as shown in Fig. 22-7.
The coils in Fig. 22-7 differ from those in a lap winding
only in t he sh ape of the leads. The diagram shows all t he
connections between the coils of phase A. In forming this
phase, let us start at coil 1. During the first tour, the
finish of coil 1 is connected to the start of coil 19 whi ch is

Ch, 22 Windings for A.G. Machines


displaced from coil 1 by 21: = 2 X 9 = 18 slots or an angle

,,?p = 2n/p = 1800 One complete passage round the core
will encompass p coils. Thus, within one tour, a given
phase contains p coils connected series-aiding. In our case,
one tour encompases two coils.
The first tour or wave must be followed by the second,
third, etc., to give a total of q waves in the same direction.
For the winding not to close upon its elf at the end of the
passage, the spacing between the last coil in a previous wave
(say , coil 19 in the first wave) and t he firs t coil of the next
wave (say, coil 2 of the second wave) must be' 21: -+ 1, rather
than 21:. For the arrangement in Fig. 22-7, this spacing is


2 X 9

+1 =


On completing q waves, we shall have obtained the first

part of the winding, Al- Xl, containing in our example
q = 3 waves (the first wave consists of coils 1 and 19, the
second of coils 2 and 20, and the thir d of coils 3 and 21).
The second part of the winding is formed in a similar manner, starting at coil 10 which is displaced from coil 1 by
1: '= 9 slots or on angle ,,?p/2 = 900 The start A2 of the
second part of the winding will be the start of coil 10 , and
its finish X2 will be that of coil 30.
The two parts of the winding are perfectly identical,
because they have the same number of coils connected in
series ai ding (each part contains pq coils) . 'I'herefore, they
ma y be conn ected not only in series, but in parallel as well.
When connecte d in series, they form a winding with one
path (or circuit) per phase (a = 1). When connected in parallel, they form a winding with two paths (a = 2).
When a = 1 (see Fig. 22-7), for t he currents to flow in
the chosen posi t ive directions, t he two par ts of the winding
must be connected in oppo si tion , that is , the finish Xl of
the firs t part must be connected t o t he finish X2 of t he second par t by a jumper. The star t A l of t he first part is the
phase start A , and the start A 2 of t he second part is the
phas e finish X .
When a = 2, t he two parts are connect ed in parallel.
The phase start A is connected to the start Al of the first
part and the finish X2 of the second part; the phase finish X
is connected to the finish X l of the first part and the start A2
of the second part.


Part Two. En ergy Conversi on by Electrical Machin es

The other two phases (Fig. 22-7 shows only the first and
last coil s) are formed in a similar wa y . The number of turns
per path is foun d, as before, from Eq. (22-5).

The Selection of a W inding Type

and W indin9 Characteristics

If a lap wind ing and a wave win din g have the same number

of coils (wound with wires of the same cross-sectional area,

with the same number of turns per coil We and with the same
coil span y), the same number of phases In and the same
number of paths (or circuits) a, and intended t o generate
a magnetic field with the same number of pole pairs p,
they will be fully identical electromagnetically, because,
given the same current , the phases set up identical magnetic
fields. They onl y diff er in the total length of wire required
t o make coils and coil conn ections. With a la rge numb er of
tu rn s per coil and a lar ge number of slots per pol e per ph ase,
the effect of coil ends is insignificant, and the total length
of wir e is practically t he same in either case. With a small
number of slots per pol e per phase, q ~ 2 or 3, a large number of pole pairs, an d a small number of turns per coil,
especially when We = 1, a wave winding is mor e attractive.
Then the saving in con duct or material may be as high as
5 % to 10 %. The larger figure applies to ma chi nes with
a relative core length equa l t o ll : ~ 1. 5, and t he smaller
figure to machines with a rela tive core length equal to ab out
3.0 . With a sm all numb er of pole pairs (say, p = 1 or 2),
wher e t he len gth of jumpers between the two parts of a ph ase
winding (A I -Xl and A2-X2) is large in com par ison wi th I
the to t al length of coil ends , t he use of a wave wind in g \
offers no advantages.
Practically, a single-turn coil is made by soldering,
brazing or wel ding together two halves, called bars.
Win dings in whi ch all t he coil sides carry t he same current
Iw e/a = the sam e
are equal as regards the pr odu ction of a magnetic fi eld. At
the same t ime, t hey may have a differen t number of turns
per coil, We, an d a different numb er of pa ths (or circuits)
per phase, a . As an exa mple, Fi g. 22-8 shows the coil sides of
three windings identical in terms of the magnetic field produced. (It is assumed that in all other respects these windings


Ch. 22 Windings for A.C. Machines

do not differ from one another. Notably, they have the same
number of slots , the same number of pole pairs, the same
number of phases, the same coil span, and the sam e current
density in the conductors .)
When the current per coil side is the same, the numher of
cir cui ts and the number of turns are chosen t o su it the reliability requirements and to simplify t he manufacture. If





l ..;

fWc ~



a: (e)

Fig. 22 -8 Windings with the same current per coil si de, I wela:
I - active conductor carryi ng f la; 2-turn insulation betwe en act ive
conducto rs; 3-ground insulation

the number of circuits can be chosen such that the resultant

coil will be a single-turn one, We = I , it is usual to pick the
arrangement shown in Fig . 22-8c, because in a single-turn,
two-layer coil (usually made of bars soldered, brazed or
weld ed tog ether), t he ground insulation also doubles as t he
turn insulation . As a result, its manufacture requires
a smaller quantity of insulating materials, and the coil
takes up a sm aller space in the slot , whereas the overall reliability is markedly improved . For a comparison of t he quantity of insulation require d, see Fi gs. 22-8a and b. '

A Two-Pole Model of a Win ding.

!Electrical Angles between W inding

If we go roun d t he peripher y of a multipole, polyphase winding, we shall see that its structure is a repetition of some
basic patternvand this repetition occurs in .an angular
dist ance equ al t o a pole pitch. For example, the pat-

Part Two. Ene rgy Conversion by Electrical Machi nes


tern tha t th e winding in Fig. 22-3 has betw een slots 1

and 18 is fully repeated between slots 19 an d 36 (in
going coun t er-clockwise from pole to pole). In such a windin g, the cur rent s in the slots displaced from one another by
2'tk ar e
a whole number k of pole pairs and numb ered N



Fi g . 22-9 Two-pole models of double-layer windings

= 3, m

(a) for the winding in Fig. 22-3 (1' = 9, y = 7, q

a win din g wit h r = Y = 9, q = 3, m = 3

3); (b) for

always the sam e. For exam ple, in slo ts N = 6 and N

2ruk = 6
2 X 9 X 1 = 24, the currents of ph ases C
and B are of the same magnitude and flowing in the same
directio n.
To form a complete idea about the winding structure, it
suffices to consider its pattern between an y pair of adjacent
poles. Consi deration of its structure between other pair s
of adj acent poles will add nothing new to our knowledge.
A t wo-pole mod el of a multipole winding is the winding
of a t wo-pole machine h aving the same number In of phases,
t he same number q of poles per pole per phase, the same pole
pi tch 't, t he same coil sp an y, and the sam e number of turns
per coil, WlJ '
A two-pole model of the winding in Fig. 22-3 appears
in Fi g. 22-9. It is an easy matter to see that it h as the same


Ch, 22 Windings for A.C. Machines

structure as the winding shown in Fig. 22-3 between an y

pair of adjacent poles. (The currents in slot N of the model
2'tk of the prototype
are the same as t he currents in slot N
winding, where k = 0 or 1, and 't = 9.) The model equally
applies to any pair of adjacent pol es (that is , between slots
1 and 18, or betwe en slots 19 and 36).
From acorn parison of the prototype winding shown in
Fig. 22-3, with its mo del shown in Fig. 22-9a, it can be seen
that the angular period of the model, ex p = 2n, is p times
as great as that of t he prototype winding given by



exp = PYp = 2n
Th e angles between the winding elements in the model increase by the same factor as compared with the corresponding
angles in the prototype

ex = PY

where Y = angle between some elements of the prototype

ex = angle between the same elements in the model.
In the theory of electrical machines, the angle y between
some elements in t he prototype is referred to as the mechanical angle*, whereas the angle ex = py between the corresponding elements in the two-pole model is called the electrical
The electrical angles bet ween the characteristic elements
of a winding (the angles in the model) determine the
fundamental properties of the winding (irrespective of the
number of pole pairs on it). The angular period of a 2p-pole
winding corresponds to an electrical angle


= 2n

The too th (or slot ) pitch in a model spans an angle

ex z

= 2n /2mq = nlmq


In a multipole win ding, one t oot h pitch spans an angle


2n /Z

2n/2pmq = nlpmq

'" In this te xt , it is referred to simply as the


angle. ~ Tra nsla to r'$


Part Two. Ene rg y Conve rsion by Electrical Machines

In the model, the phase belt occupied by t he phase conductors (in one layer) wi thin each pole pi tch spans an angle
that is, '1/m part of a pole pitch . (In a multipole winding,
this belt spans an angle Yq = qyZ = nl pm.)
The coil in a model spans an angle



In a multipole winding, the coil spans an ang le


(y!T:) (Yp/2)

(yh) (nip)

The procedure for deve lop ing a two-pole winding model

does not differ from that set forth above for a multipole
winding. As an example , Fig. 22-9b shows a mo del winding
which differs from that in Fig. 22-3 onl y in having a full
pitch, that is, y = 1:". As is seen, when y = 1:" , the top and
bottom l ayers of the winding are not disp laced from each
other, so that all the pha se cond uct ors within a given pole
pitch are only laid in q slo ts (in our case, q = 3). Comp ar e
this with Fig. 22-9a, where y = 7, 1:" = 9, and the phase
conductors within a pole pitch ar e la id in q
= 3 9 - 7 = 5 slots.


Two-Layer, Fractional-Slot
W indings

The two-layer . win din gs exa mi ned above are integra l-slot
wind ings. This means t hat they h ave an int egr al number q of
slots per pole per phase, and t he number q of coils in each
phase remains the same from pole pitch to pole pitch (in
Figs. 22-3, 22-5, and 22-7, q = 3 coil s in each ph ase). This
is the most commo nl y used vari et y of tw o-layer win ding.
Ano ther is wha t is calle d the fra ct ional-slot tw o- layer wi ndi ng .
In such a winding, the poles of, say, the rotor are design ed
to occupy only a part (fraction) of the st a tor sector t hat bounds
three slots (in the case of three-ph ase machi nes) of the stator
winding , or less t han one slot per ph ase per pole. Th a t
is why t hey are call ed fract iona l-slot win dings. With a small
number of slots per pol e per phase (q < 3) an d a large number
of pole pairs , such windings offer a numb er of ad vantages
over integral-slot windings.

Ch, 22 Windings for A.G. Machines


To obtain a fractional-slot winding, it is necessary that

there be one coil more in some phase coil groups than in
others. Those with one coil more (the l arger or major phase
groups) will then have (b
1) coils each, and the others
(the smaller or minor phase groups) will have b coils each.
Becau se of this, the winding does not r epeat itself every
pole pitch , but does so in a basic pole or coil pattern frequ ently called the r epeatable (pole or coil) group . Denoting the number of maj or ph ase groups in one basic pattern
as c and the numb er of m ajor and minor phase groups (or
poles) in one basic pattern as d, the number of slots per pole
per phase for fractional-slot winding may be written

q= b

+ cld

As a rule, use is mad e of fractional-slot windings with

b ~ 1. In them, each phase has 2p coil groups (one group '
per pole). Of this numb er , the m ajor groups will be under
n = 2p/(c/d) poles, and the minor groups under (2p - n)
poles. Then the number of slots per pole per phase for a
fractional-slot winding ma y be written

q = n (b+1 )+(2p-n) b =b+ c/d


As alread y noted, the denominator of t he fraction is

the number of pole pitches in one basic pattern . In a com. pl ete winding, there may be 2 pld such basic patterns. For
a winding to be feasible, 2 pld must be an integer. A further requirements is th at in a symmetrical pol yphase winding the denomina tor should not be a multiple of the number of ph ases. Since th e winding . must produce a periodic
field , it mu st have an even numb er of poles. Ther efor e,
the repea table group in a fr actional-slot winding t ak es up
d pole pitches wh en d is even and 2d pol e pitches when d
is odd .
When d is even, t he winding h as 2 pld basic patterns;
when d is odd , it h as 2p/2d basic pattern s.
To sum up , a fractiona l-slot winding h as "m inor" phase
groups of b coils each and "m ajor" ph ase gr oups of b
coils each . These groups alternate in a sequence whi ch depends on the m agnitude of the fra ctional part of the number q. Th e denominator is the number of all phase coil groups
in which the. sequence of major and minor groups is repeated. One s equence made up of d coil groups contains d - c


Part Two. Energy Conversion by Electrical Machines

minor groups and e major groups. Each pole pitch corresponds to one phase group (with q > 1). In a three-phase
winding the total number of phase groups is 6p , so th e
sequence rep eats itself 6 pld. times. When b = 0, the winding will consist solely of major groups, with one coil each.
The maximum number of circuits (paths) in a phase winding is a max = 2pld. The lowes t possib le number of circuits
is such that 2p lad is an integer.
.A simple procedure to construct a frac tional-slot, twolayer lap winding is as follows .
1. Det ermine the number of coils in a minor group, b, and
in a ma jor group , b
2. Write a series of e numbers: die , 2dle, 3dle, ... ,



Re place each fractional number by the nearest integer

number so as to obtain a ser ies of e numbers: N I , N 2' N 3'
... , d. These are the Nos. of ma jor coil groups arranged in
the same sequence as the coil groups of all the phases are
arranged around the core periphery for one repeat able group .
3. Assign numbers 1, 2, . . "., N I - 1 to the minor coil
groups of b coils each. The next N 1 th coil group, a major
1 coils. In a similar way, form the other
one, consists of b
coil groups, assigning numbers N 2 ' N 3 ' . . , d to major
groups; the rem aining groups will be minor ones. Follow the
same sequence in forming the rem aining 6pld - 1 rep eatabl e
groups (a three-phase winding is meant) .
4. Distribute t he coil groups am ong the phases. E ach of
the phases ta kes every third coil gr oup .
Choose the posi tive direct ion s of currents so t hat in ad jacent )
coil groups (belonging t o different ph ases) t hey are in
oppo site senses. Connect in series opposition t he adj acent
coil groups in each phase.
5. One pa th (or circuit) can be formed from d seri esconnected phase coil groups.
6. Once the coils of phase A hav e been connected (the
ph ase start can conveniently be combined with that of the
first coil group in the firs t repeatable group), the leads and
connections between the paths in phases Band C may be
chosen in anyone of several ways, namely:
(i) all connections between the coil groups in phase B
may be made similar to those j n ph ase A , with a displac ement equal to two coil groups. The n the connections in

Ch, 22 Windings for A.C. Machines


phase C will repeat those in phase A, with a displacement equal to four coil groups;
(ii) all connections between the coil groups in phase B (C)
are displaced from the connections in phase A by one basic
pattern (that is, by d coil groups) if d is even, or by two
basic patterns (that is, by 2d coil groups) if d is odd. In phase
C (B), the connections should be displaced from those in
phase B (C) again by as many coil groups.
In case (i), the phases are identical only as regards the
production of the magnetic field; the sequence in which the
minor and major coil groups are connected in the phases is
different. In case (ii), the phases are nearly identical in
regard to both the generation of the magnetic field and the
sequence of coil groups. (Phase B can be formed by
turning phase A through an appropriate angle.)
Example 22-1. Given: Z = 42, 2p = 8, In = 3, q
(b = 1, e = 3, d = 4), y = 4, a = 1.
The number of coil groups in the basic pattern is

13 / 4

3d = 3 X 4 = 12

The number of coil groups in the basic pattern per phase

is d = 4.
Each minor group has b = 1 coil. Each major group has
1 = 2 coils.
The major coil groups in one basic pattern are numbered



F/ 3 , 2dle = 2 2/ 3" 3dle

Rounding them off to the nearest integers, we get: 2, 3, 4.

The number of alternations in the entire winding is
6pld = 6 X 4 --;- 4

= 6

The manner in which the coil groups are distributed along

the core periphery is as follows (the numeral indicates the
number of coils in a coil group; the vertical spaces separate
the basic patterns):






The letters A, Band C mark the forward coil groups, and

the letters X, Y, and Z, the backward coil groups in the respec-


Part Two. En ergy Conversion by Electrical Machin es

tive phases. The starts and finishes of the first basic phase
patterns, labelled as advised in (ii) above, are underscored
once ; the starts and finishes of the second basic phase
patterns are underscored twice (the star t of a basic pattern

Fig. 22-10 Fractional-slot , double-layer lap winding

= 42, p = 4, In = 3, y = 4, a = 1, q = 1 3 / 4 )


is the start of a forward coil group, and the finish of a basic

pattern is the finish of a backward coil group) .
Schematically, the resultant winding is shown in
Fig. 22-10. As is seen, the pattern of the winding (and
that of the fie ld established by t he currents) per iodically
repe ats itself, its "period" being equal t o the length of the
basic pattern which spans 2p/k pole pitches or an angle
2n/k, where k is the greatest common divisor for Z and p .
In our example, Z = 42, p = 4, and k = 2. Therefore, the


Gh. 22 Windings for A.C. Machines

repeatable group repeats itself every 2plk = 2 X 412 = 4 pole

pitches, 2pr;lk = Zpmqlk = 2 X 4 X 3 X 131 4 -;- 2 = 21
slot pitches, or ever y 2nlk = 180. In Fig. 22-10, the pattern
of currents and coils in the belt extending from slot 1 to slot
22 is fully repeated in the belt extending from slot 22 to
slot 42.
In a frac tional-slot winding, the pattern of currents (and
of the resultant field) is especiall y clearly seen to be recurring
about every two pole pitches, 2r;, or in an angle equal to
"(p = 2nlp. For example, in phase A the groups of coil sides
carrying currents flowing in alternate directions recur in
about an angle "(p12 = (2nlp)/2 = (2n /4)/2 = 45, as is
shown in the figure . The group of coil sides carrying currents
flowing in the same direction recur in the angle "(po As is
seen, these groups do not con tain an exactly same number of
coil sid es (three sides ar e laid in slots 41,42, and 1, three sides
in slots 5,6, an d 7, four sides in slots 10,11,12, four sides
in slots 15,16, and 17, etc. (In the general case, the number
of sides is either 2 (b
1) or 2b
1.) Yet, the pattern
of currents is periodic enough for the production of a magnetic
field with the desired number of pole pairs p.
No two-pole mod el can be built for a fractional-slot winding. A model representing a complete cycle of change
in the pattern must contain 2p' = 2plk pole pitches, that
is , as man y as is occupied by the basic pattern.




Field W indings

The function of a field winding is to set up a heteropolar

exciting magnetic field. It is a single-phase, heteropolar winding energized by direct current.
The two basic designs for t his winding have been examined in Sec. 19-2. Figure 19-2c shows the arrangement of
a concentrated field winding used on salient-pole cores. It
is fabricated in the same manner as a single-phase, two-layer
winding, but has only one coil per group, so t hat the number of coils per pole per phase is q = 1. Also, this is a fullpitch winding (Yc = r), and its coil sides are laid in slots
next to each other, taking up a half slot width each (see
Fig. 19-2c). The arrangement of a concentrated, two-layer
field winding laid in slo ts 1 through 8 between poles is shown


Part Two. Energy Conversion by Electric al Machines

in Fig. 22-11. The construction of a rotor carrying this type

of winding is discussed in Sec. 51-3.
In a concentrated single-circuit fi eld winding the number
of series-connected turns is io = 2pll'c' where Wc is t he number of t ur ns per coil.

Fig. 22-11 Concentrated field winding (p = 4, .. = y , q = 2)

Fig. 22-12 Distributed single-laye r

field winding (p

1, q


An alternate design of the fi eld win ding is shown in

Fig. 19-2d. This is a distributed fi eld wind ing, and it is
used on round (cylindrical ) cores. In con trast to a concentrated field winding which t ak es up only one slot within each
pole pitch (q = 1), this winding is laid in s > 1 slo ts per

Ch. 23 Calculation of Magnetic Field


pole, and each slot receives only one coil side. Therefore,
it may be treated as a single-layer winding.
As a rule, the winding has q slots per pole per phase.
Accordingly, within each pole pitch there are q/2 concentrically arranged coils. As is seen from Fig. 22-12, the coilsaof a
distributed field winding differ in pitch. For design purposes
the coil pitch in such a winding is equal to the pole pitch,
Yc = ... To make the magnetic field set up by the winding as
nearly sinusoidal as practicable, the slots carrying conductors occupy 2/3 of a pole pitch. For example, the winding in
Fig. 22-12 is laid out in 12 slots. The construction of the
rotor carrying a distributed field winding is examined in
Sec. 51-4.
In a single-circuit, single-layer distributed field winding,
the number of series-connected turns is w = pqw c ' where We
is the turns per coil.


Calculation of the Magnetic Field

in an Electrical Machine


The Statement of the Problem

Energy conversion in an electrical machine operating by

electromagnetic induction is based on its magnetic field.
Therefore, the calculation of the magnetic field established
by the currents flowing in the machine's windings is a major
problem in the theory of electrical machines.
In the general case, the problem reduces to finding the
magnetic induction (magnetic flux density) B from the
specified density current J in the windings of the machine
(Fig. 23-1), and it can be solved by the theory of the electromagnetic field.
The magnetic field strength (magnetic intensity) vector
must satisfy Maxwell's first equation
curl H = J


the equation connecting the magnetic induction and the

magnetic field strength


Part Two. Energy Conversion by Electrical Machin es

where ~L a is the absolute perme ability of the medium, and

the continuity equation
div B = 0
implyin g that t he lines of magn etic flux ar e always
In most cases, the current density vector J is uniformly
distribut ed over th e cross-sectional area Q of a conductor
J = I/Q

and points along the axis of the conduct or in the dir ection
where the current I is flowi ng (see Fi g. 23-1).

Fig. 23-1 Produc ti on of th e magneti c field

Ordinarily, the winding conductors are laid in slots on

the stator and rotor cores, and the magnetic field exists in
a space taken up by the two cores, in the nonmagnetic gap
separat ing them , and around the coil ends (or overhangs)
(Fig. 23-2). In many cases, this field even threads the magnetic and conducting structural parts of the machine (the
shaft, frame, end-shields, and so on).
To be able to calculate the magnetic field , the general
field equations (23-1) through (23-3) should be extended to
include the equations

(x, y, z) = 0


describing the surfaces separating the media i and If. differing in relative permeabilit y , /-Ir. i =1= ~r. l, (above all , the


Ch. 23 Calculation of Magnetic Field

equations describing the surfaces bounding the cores), and

also boundary conditions for the tangential and normal
components of the magnetic Iield vectors at the surfaces

Fig. 23-2 Magnetic field set up by a basic coil set

separating the ith medium from the kth medium


Ht;i = Ht,k


/-Lr,i Hn, i =

/-Lr,h H




tangential components of the magnetic

field strength on the boundary
En, i, En , l< = normal components of the magnetic flux
density at the same points on the houndary.
In cases where the permeability of the cores, /-Li',C' cannot
be deemed infinitely large in comparison with that of the
areas taken up by air, insulating materials , and winding
conductors, it is essential to take into account the nonlinear
magnetic properties of the ferromagnetic materials, that
is, the dependence of the relative permeability on the magne17*

Ht,i' Ht,l< =


Part Two. Energy Conversion by Electrical Machines

tic field strength

fLr,c =

f (II)


Equations (23-1) through (23-6) uniquely describe the

magnetic field in a machine, but they cannot in most cases
be solv ed analytically by electromagnetic field theory.
This is, above all, because the surfaces bounding the cores
and current-carrying conductors are intricate in shape and
also because one would have to consider the nonlinear mag- .
netic properties of the ferromagnetic materials. Further
difficulties arise because the relative position of the cores
and current-carrying conductors is changing all the time,
and the solution would have to be sought for all the likely

Assumptions Made in Calculating

the M agnetic Field

rn the theory of electrical machines, several simplifying

assum pt ions are made so that the magnetic field of a machine
and the relevant winding characteristics could be determined analytically.
1. Because the current pattern in the windings repeats
itself periodically (see Sec. 22-2), the field pattern is likewise repeated periodically every two poles. Therefore, in
calculating the magnetic field of a machine, it will suffice
to consider its variations over a pole-pitch angle 'Vp or even
over a half of the pole -pitch angle, 'Vp/2 . On an enlarged
scale, Fig. 23-3 shows the magnetic field over a half of a pole
pitch in the machine shown in Fig. 23-2.
2. It is assumed that the ferromagnetic cores have an
infinitely large permeability, fLr,c' in comparison with that
of free space . Because at a magnetic induction of 1.5 to 2.0 T
the relative permeability of the core is several tens or hundreds , this assumption does not introduce any appreciable
error in the calculation of the field. Also, one may allow for
the finite value of fLa,c and of the reluctance of the ferromagnetic parts of the magnetic circuit at a later stage, in
practical calculations.
3. Once the assumption in (2) above is made, we may use
the principle of superposition and treat the magnetic field
of the machine as the sum of the fields set up by each of its
windings. In turn, the field due to a winding may be treated

Ch, 23 Calculat ion of Magne tic Field


as the sum of the fields established by basic or repeatable

coil sets* .
The term "basic coil set" refers to a set of 2p coils un iformly distributed all t he way round the circle , disp laced from
one another by a half pole pitch , and in terconn ect ed so as
to form a per iodic current pattern with p pole-pairs.

Fig . 23-3 Magneti c field from Fig. 23-2 show n enlarged withi n a polo

In Figs . 22-3, 22-5, and 22-7, such a ba sic coil set is

shown by heavy lines. In these figures, the to p coil sides are
laid in slots 1 , 10, 19, and 28. Referring to these figures, it
is an easy matter to see t hat any winding may be decomposed into a multiplicity of basic coil sets. A winding phase
is formed by q basic coil sets displaced from one another in
space by one tooth pitch . The entire winding has mq such
basi c coil sets .

* We

consider only integral-slot win dings (see Sec . 22-2).


Part Two. Energy Conversion by Electrical Machines

To determine the total field, it will suffice to calculate

the magnetic field due to one basic coil set carrying a unit
current, I = 1, to find the fields of all the basic coil sets
by scaling up the field due to a unit current, and to combine
these fields subject to their relative position in space.
In this way, the problem of finding the magnetic field set
up by the currents in all the windings of a machine reduces
to calculating the magnetic field established by a basic coil
set repeated periodically in the winding structure, assuming
that the cores have a relative permeability of infinity.

The Spatial Pattern of the Magnetic

Field Set Up by a Polyphase Winding

An idea about the spatial pattern of the magnetic field established by a polyphase winding can be gleaned from reference
to the field established by the basic coil set (see Fig. 23-2).
This is a four-pole field. Each two-pole interval contains
one short-pitched coil. In moving through a half the polepitch angle, ,,?p/2, the field pattern repeats itself, but with
the signs reversed. Assuming that the permeability of the
cores is infinitely large, the magnetic field within the cores
need not be considered because its energy is zero (in Fig. 23-3
it is shown by dashed lines). In the nonmagnetic areas, the
magnetic field may be decomposed into three components,
(i) the field in the air gap
(ii) the field in the wound slots
(iii) the field around the coil ends (overhangs).
In machine design, the most important factor is the air
gap field , that is, the field in the clearance between the
cores. In terms of energy, this field exceeds the other flux
by a wide margin , which is why we shall give it most of the
treatment in our further discussion. In Fig. 23-3, the flux
lines in the air gap are shown by solid and heavier lines.
The distinctions of this field may be summed up as follows.
Firstly, within the core length l its lines lie in planes at
right angles to the z-axis, and the flux pattern repeats itself
in each of these planes, so we may call it a planar (or twodimensional) field. Secondly, all flux lines cross the air
gap and determine the flux linkage and mutual inductance
between the winding in question and the windings laid on
the other core, for which reason we may call it a mutual field.

Ch. 23 Calculation of Magnetic Field


Thirdly, no distributed currents exist within the region

taken up by this field, so in calculating it we may invoke the
concept of a scalar magnetic potential (see Sec. 23-4).
On an enlarged scale, the field in a wound slot, that is,
one enclosing current-carrying conductors, is shown in
Fig. 23-4. Its lines link only with the conductors of the
winding in question. They never cross the air gap, nor do
they link with the windings laid on the other core. Such
field ( :are called leokagejields.
The region taken up by the
slot leakage field is separated from that occupied by the
mutual air-gap field
characteristic field lines 01 and
04 which pass through point
o on the surface of the other
On a closer examination, the
slot leakage field is seen to
be the sum of a leakage field
in the slot (that is, one existing
inside the slot as far as line
23), and a leakage field in the
tooth, whose lines extend into
If the air gap and exist within
the region 012340.
Within the core length, the
slot leakage field is planar.
Fig. 23-4 Enlarged element of Its pattern repeats itself at
the magnetic field in Fig. 23-2,
any section of the machine and
around a current-carrying slots
its lines lie in the section
planes. The slot leakage field
is more difficult to calculate than the air gap field, because of
the distributed current existing in the slot region. (The current density J within the cross-section of a coil may be
taken constant and directed along the z-axis of the machine.)
In the general case, the slot leakage field can be found,
using a general description of a magnetostatic field [see
Eqs. (23-1) through (23-5)1. Still, although the field calculation is materially simplified because the field is two-dimensional, an analytical solution can only be obtained for some
particular cases (say, for rectangular or circular slots).
Even then the analytical solution is too unwieldy, and


Part Two. Energy Conversion by Electrical Machinos

practical calculations for slots of any shape are based on the

approximate solutions deduced by idealizing the field pattern. Such an approximate solution will be considered in
connection with the slot leakage inductance (see Sec. 28-7).
The coil-end (overhang) field refers to that around the coilend connections and outside the cores. Its lines are closed
around the coil ends and form a complex spatial pattern.
For the coil-end field to be determined accurately, one would
have to use a complete description of the magnetostatic
field such as set forth at the beginning of this section. The
solution of the problem is complicated by the fact that the
field is three-dimensional and the coil overhangs are very
complex in shape. Also, the field may to some extent be
affected by the ferromagnetic parts of the machine, such as
the end shields, frame, shaft, etc. On the other hand, we
are free to neglect the angular position of the rotor relative
to the stator.
The coil-end field is very low in energy. Therefore, the
accuracy in calculating the coil-end field and inductance
(see Sec. 8-7) need not be very exacting. It is important to
note that some lines of the coil-end field link with the coil
ends of the windings on the other core and contribute to the
mutual inductance between the Windings. Therefore, the
coil-end field proper refers only to a fraction of the total
coil-end field. The contribution of the coil-end field to the
mutual inductance between the windings on different cores
is very small (in comparison with the effect produced by the
mutual air-gap field) and may be safely ignored.

Calculation of the Mutual Magnetic Field

for a Polyphase Winding

On the assumption made in Sec . 23-2, the mutual field of

a polyphase winding is planar (two-dimensional), and its
energy is concentrated in the air gap where distributed
currents are non-existent. Its strength 11 may be expressed
as the gradient of a scalar magnetic potential, qJm = qJ
11 = - grad 'P


On substituting Eq. (23-7) into Eqs. (23-2) and (23-3), we

can readily obtain for the scalar magnetic potential a secondorder p.artia~ differential e~uation/ known as the Laplace

Ch. 23 Calculation of Magnetic Field


equ ation:
To determine cp at any point (x, y) in the air gap, we must
solve Eq. (23-8) subject to the boundary conditions corresponding to the instantaneous currents in the winding phases
and existing on the ferromagnetic surfaces. The boundary
conditions are specified by giving the dis tribution of the
potential (p on the surfaces . The determination of this distribution is a problem in its own right, and it can be solved
unambiguously, if we know the winding circuit and the
instantaneous currents in the phases . Obviously, the solution
becomes progressively more difficult to obtain as the winding
grows more complicated in arrangement. Therefore, it" is
advantageous to solve the problem first for the currents in
one hasic coil set, and then to find the potential distribution
for a polyphase winding by adding together the potentials of
all the basic coil sets.
With cp found by solving Eq. (23-8), the components of
the air gap field are found hy Eq. (23-7):
H x = - ocp/ox,

= - ocp/fJy


Effective Length of the Core

Figure 23-5 shows the machine of Fig. 23-2 cut lengthwise.

One of the cores is divided into several pa ckets of length Z;
each, separated by radial cooling ducts of width bd As is
seen from the gap field pattern, the air gap field is nearly
uniform and constant in the region taken up by the core
packets (in fact, within the cross-section passed through
this region the fi eld ma y be regarded as pl anar); is somewhat
weakened in the ducts, and gradually collapses on emerging
from the core faces and on leaving the air gap.
All this has a well-defined effect on the distribution of the
radial (normal) components of the air gap flux density, B.
To sim plify fur ther cal culations wi thout mistreating energy
conversion hy the machine , we may replace the field varying
along the length of the machine hy a uniform field with
a maximum flux density B m in the packets. In doing so, we
also assume that this uniform field exists over the design


Part Two. Energy Conversion by Electrical Machin es

or effective core length lo such that


= .\' B dz = B ml o


lo =

- '1 -


J' B dz

It can be shown that


where bd =
Cd =
Co =
Co =

(b d/co6)/(5
b d/co6)
1, if ducts are made in the stator (or rotor) only
0.5, if ducts are made in both the stator and
rot or .

Fig . 23-5 Det ermining the design length of the core

Also, if the ga p is very small (6 ~ b d ), the design or

effect ive width of a duet is b;l ~ b d If the gap is ver y l arge
(8 ~ b d ), the design duct wi dth is ba ~ O.
It may be added that, as oft en as not , lo stan ds for the
axial gap length ,


Ch, 24 Mutual Magn etic Fi eld of Phase Winding


The Mutual Magnetic Field

of a Phase Winding and Its Elem ents


The Magnetic Field and MMF Due

to a Basic Set of Currents

A basic set of currents periodically alternating in dire ction

every t wo pole pit ches, 21:, is shown in Fig. 24-1. The currents
iw c and -iwc are carried in slots on core 'Cl . A displacement through a pole pitch, 1:, causes the direction of current
flow in a slot to reverse. The excited core Cl is separat ed




iw c

Fi g. 24-1 Repeat able. pattern of currents

from the unexcited core C2 by an air gap of wid th tJ . Because

it is sma ll in comparison with R , the mean radius of t he
air gap , we ma y neglect t he effect of t he curvature and replace
the annular air gap by a "develope d" or fl attened- out gap
(see Fig. 24-1). A reference point in the developed gap can
conveniently be located by t he dis t ance x from the slot ax is,
whi ch is connected to the angular coordinate of t he point in
the annula r air ga p, y, by a simple relation
x = yR
To simplify the analysis , it is advan t ageous t o replace t he
distribute d slot curren t , as shown in Fig. 24-1, say iwc , by
an equa l linear current, i s = iw c , concentra t ed at t he axi s
and nea l' the bottom of t he slot (Fig. 24-2) .

Part Two. Energy Conversion by Electrical Machines


The magnetic potential existing at the boundaries of the

air gap, coinciding with the surfaces of the cores C1 and C2
can be found by applying Ampere's circuital law to the loop
1'-2'-2-1-1' which is symmst.
rical about the slot axis:




~ Hz dl


Assuming that the permeability of the core is infinite

and the magnetic flux1f density B within the core is
finite, we may write


H = B/[La = 0
so that the magnetic potential
experiences no drop within
the cores



-twe /2
_ .


I Hz dl

Fig. 24-2 Mutual field and mmf

in and around a wound slot
(see Fig. 24-1)

tt.ei = 0



Therefore, the circulation of

the vector II may be written as
the sum of variations in the air gap potential within portions 1'-2' and 2-1:


HI dl =

Hz ell

+ .I HI dl = j


Hz ell = iur;



If we recall that the magnetic field is symmetrical about

the slot axis, so that

.\ Hz dl = \ Hz dl = iw c /2


and set equal to zero the magnetic potential of the unexcited

core, (pz = 0, then the magnetic potential of the excited core
to the right of the slot axis (with x > 0) will be

(PI ~


+ J Hz dl = iw,)2


Ch , 24 Mutual Magnetic Field of phase Winding

and to the left of the slot axis (at x



(Pt =

CP2 -


.\ HI dl = - iw c /2

It follow s from Eq. (24-4) that the potential of the excited

core to the left of t he slot axis differs from t hat to the right
by the slot current, iw c ' (The current flowing outwards , that
is towards the reader, is t aken to be positive .)
The air gap field dep ends on the difference in magnetic
potential produced between the surfaces of the two cores by
the currents in the respective windings. In the theory of
electrical machines, this difference in magnetic potential;
equal to t he linear integral of the air gap field intensity
or the to tal air gap current, is usually called the magnetomotive [orce, or mm i for short.
Choosing as positive for the air gap field and mmf the
direction away from the inner core, C1, t owards the outer
core, C2, we may define the mmf as

F = CPl - CP2
In our case, C1 is excite d, C2 is unexcited , and
the mmf of C1 is
F 1 = CPl - CP2 = CPl
If C2 were ecxited and C1 unexcited , and CPl
the excited core would be

(P 2

= 0, so

= 0, the mmf of

F 2 = CPl - CP2 = - CP2

Wi th the boundary conditions given by Eqs. (24-3) and
(24-4), the magnetic poten tial in t he air gap can be found by
Eq . (23-8). If the potential distribution is known, t he field
intensity can be found by Eq . (23-9) . At some distan ce from
a slot, however, t he magnetic field strength can be found
in a simpler way. As follows from Fig. 24-1 , the lines of the
magneti c field are complex in sh ape only near a slo t, wher eas
at some distance from the slot, I x I > 5, the field becom es
practically uniform ; its lines run normal t o the core surfac e,
and it s in tensity is t he sam e at all t he poi nts wi thin the
air gap.
Choosin g the path of integration (1'-2' or 1-2) t o run along
a field line (where the field is uniform) and noting that

Hz = H y = H



Part Two, Energy Conversion by Electrical Machines

we get

Fi=CPmi -CPm2=

HI dl =

JH yd y =H8

The magnetic field strength in the gap is

= F l /5 = Fllv

where Iv = 1/5 is the permeance of the air gap within a region

containing a uniform magnetic fi eld.
The magnetic flux density in the air gap is given by
In an electrical machine with a saturated core , the mmf is
a sum of several components each of which balances the
magnetic potential difference within a certain por tion of
the magnetic circuit. These component mmfs are found by

The Effect of Core Saliency.

The Carter Coefficient

Figure 24-3 shows t he magnetic field set up by a basic periodically repeatable set of currents, iw c and -iwc , carried in
some of the slots of core Cl. The figure shows one slot carry ing iw c (a wound core) and several slots carrying no current (unwound cores). The slot width b; is assumed to be
comparable with the gap width 5. Then the field in the
region of the unwound slots is markedly reduced, and its
strength is substantially smaller than it is in the teeth .
The magnetic flux across the air gap can be expressed in
terms of the normal component of the air gap field intensit y,
fln = H y ' For example, the flux across the area bounded by
tooth pitch 3-4 is

cD 34 =


l-toHn dA =

Jl-toHyl~ dz

where l o is the effective core (or axial gap) length .

In many cases, however, one need not know t he exact
dis tribution of the normal field component over each tooth
pitch . Instead, one may limit oneself to the distribution of

Ch, 24 Mutual Magnetic Field of Phase Winding


the mean normal component, H o' which is taken to be such

that the magnetic flux across the area bounded by a tooth
pitch (unwound) remains unaffected:

H0 =

(J)3 4//100tZl {) =

J H y dx/t z


A detailed study into the air gap field will show that
when the toothed core Cl is replaced by a smooth surface


Fi g . 24-3 The effect of core salie ncy on the magneti c fiel d nea r an
unwound slot

separated from C2 by a distance 6 0 >6 , the mean normal

component of the air gap field, H o' will remain the same as
it was with a toothed core, provided
60 = 6k{)
Here, k () is the Carter coefficient, named the airgap factor
in t he USSR . When the actual air gap 0 is multiplied
by k (), the product gives the effective gap width, 0 0 , . The
air gap factor is given by
.' k"o = tz/(tz .- cso)


Part Two. Energy Conversion by Electrical Machines


= (bs/6)2/(5

+ bs/5)

In cases where both C1 and C2 are toothed, the effect of

their teeth can be accounted for by applying the compound
air gap factor
k 6 = k 6 1k 0 2

where k 01 and k 62 are the air gap factors of C1 and C2,

respectively. Each is found by Eq . (24-9), assuming that the
other core has a smooth surface:
k 01

t z 1/ (t Zl -

= (bS l/6)2/(5

C S1

!c 0 2
CS 2


t Z 2/ (t Z 2 -

= (bS 2 /5)2/(5

cs 15)


S l/8)

c s 2 5)

+ bS2/8)

The MMF Due to a Basic Coil Set

A basic coil set is the simplest repeatable element of

a phase winding (see Sec . 23-2) . Therefore, prior to
determining the magnetic field of a phase, we should find
the mmf due to a basic coil set carrying the phase path
(circuit) current, ia.
The instantaneous current in a phase path (circuit) is
given by
i a = i/a =

V2 I a cos (rot)


where i is the instantaneous phase current, I is the rms phase

current, and I a = IIa is the rms path current.
The pattern of currents carried by the coils in a basic set
is repeated every two poles. Therefore, it will suffice to
consider the mmf and field due to this set over two pol e
pitches, as shown in Fig. 24-4. Each pole pitch is seen to
contain one coil of the basic set . The instantaneous phase
current is assumed to be flowing in the positive direction
(that is, from its finish to its start). Its direction at the
coil sections is shown in the figure.
The mmf due to a basic coil set can be visualized as the
sum of the mmfs due to two periodically recurring sets of
currents, namely F' due to the currents in the odd-numbered
slots (1, 3, 5, etc.), and F" due to the currents in the evennumbered slots (2, 4, 6, etc.). The mmfs due to periodically


ClI. 2/1 Mutual Magn et ic Field of Phase Winding

rocurrlng set s of cur rents ha ve been defi ned in Sec. 24-1.

Graphi cally, th ey are combined in Fig. 24-4. In side a coil
pit ch, Ye,

F = F'
F " = + iawe

Between coils, F = O.
On mo vin g in the positive direction, t he mm f at the slot
axis is in crement ed by the slot cur ren t i alV e if th e cur ren t is

f ig . 24-4 MMF du e to a basic coil set

fl owin g towards t he reader , or decrem en t ed by the same

amount if the current is flowing away from the reader. The
mm f is thus seen to vary peri odically with a period equa l t o
two pole pit ches , 2 . Th erefore, a displacement of 2. leaves
the mmf with it s origina l sign
F (x

+ 2. ) = F


whereas a disp lacement of causes it to change sign

F (x

+ .) = - F


(2f [-13)

Assuming that the positive directi on for the fi eld an d the

mmf is fr om the exc ited to the u nexcited core and taking
18-01 0 0


Part Two. Energy Conversion by Electrical Machin es

as th o origin the axis of t he coil setting up a positive phase

mmf when th e cur rent is flowing in t he positive direction,

we ma y write the following equation for the mmf over one

pole pitch:
F e = ialVc = F em cos wt for -Ye/2 < x < Ye/2

for - . /2
for . /2




-Ye/2 and


Ye /2

where Fern = V2Ia lV c is the peak mmf due:to the basic coil
Thus, when the basic coil set is carrying a sinusoidally
varying current i a , the resultant mmf is a wave stationary



-oh -;


L - rL..L:~

- --(

Fig. 24-5 Br ea thi ng mrnf waveform

in space and pulsating at an angular frequency co = 2nf.

The position of the wave in space depends on the arrangement of the coils, and the magnitude of the mmf is determined by the value of i a
Figure 24-5 shows the "breathing" mmf waveform during
one cycle of change in the current. Equations (24-12)
through (24-1 4) completely describe the mmf all the way
round the periphery of the air gap having p pol e pairs. For
the ma chi ne of Fig. 22-3 in which p = 2 , the dist ri bution
of the mmf along t he periphery of the air gap is shown in
Fig. 24-6a .
The position of an ar bitra ry point in the air gap can be
specifi ed by giving eit her the distance x from the origin
along the periphery of the air gap, as indicated on the developed ("unfolded") view of the annul ar gap (see Fig . 24-4
and elsewhere), or the angle I' from the origin to the point



24 rvilltuai Magneti c Ficld of Phase Winding

in question:

'\, = xlR = sd-cp



where R = 1:pln is t he me an radius of the air gap circumference .

Because t he mmf pa ttern is rep eated every two poles,
all that is necessary to know about the mutual field in
a machine can be gleaned from its two-pole model. This
model should retain the winding arrangement, as does the

Fig. 24-6 Distribution of the mmf along the periph ery of the air gap
in (a) a 2p-pole machine and (b) its two-pole model

two-pole model of the winding (see Sec. 22-6), the slot and
t ooth dimensions along the periphery of the air gap (bs' t z ,
1:, and Yo), and also the radial gap length B and the effective
axial gap length lo.
The length of the gap circumference in a two-pole model
of the machine is 21:. As compared with t he length of the
air gap circumference in the prototype machine , it is reduced
by a factor of p. Therefore, the radius of t he air gap in the
model is likewise 1lp of that of the actual gap , th at is, Rip.
In the model, as in the prototype machine, a given point
within a pole pitch takes up a position defin ed by t he same
distance x from the origin O. The angle ex. specifying the
position of t he similar point in the model, called the electrical
angle, is p times the mechanical angle in the prototype
i 8*

Part Two. Energy Conversion by Eiectricai Machines

machine. Since
x = 'VB = o.Rlp

and subject to Eq. (24-15), it follows that

a = P'V = (aIr;) rr


In going from a proto type machine to its mod el, the angles
between any characteristic machine elements within a pole
pitch are multiplied by the same factor:
a1' = P'V]J = 2n
a y = P'Vy = ynIT:

To sum up, the electrical angle a between any two machine

elements within a pole pitch is thought of as the angle be tween the same elements in the two-pole model of the machine
in which a cycle of change in the field is completed within
an angle of 2n .

Expansion of the Periodic MMF due to

a Basic Coil Set into a Fourier Series.
The Pitch Factor

Let ui?: expand t he r" mmf due: to a repeatable coil set into
a Fourier series for t = 0, when the current in a parallel
path is a positive maximum
ia =

Then the mmf . in

will be

V 2I a

the coil region

F crn =

(-Y c/2




V 2 I aw c

The mmf waveform for t = 0 is shown in Fig. 24-7. As

is seen, the mmf is an even function about the axis passing
through the middle of the coil; therefore, the Fourier series
will only consist of cosine terms. Also, during the next halfcycle of change t he waveform repeats itself, but with its
sign reversed. For this reason , the series can only contain
odd harmonics [14]. Figure 24-7 shows the fundamental
mmf of peak value F C1 which completes a half-cycle of change
in a time equal to 't1 , and the vth harmonic mmf of peak
value F c v which completes a half-cycle of change in a time
equal to 'tv = xl ,

Gil. 24 Mu tu al Magn etic Fi eld of Ph aso Winding


Th e mmf can be represented as a sum of h armonic te rms


Ft=o = ~ F c v m cos (v xlo:) :rt


v= 1

where v = 1 + 2c = 1, 3, 5, 7, etc ., an d c = 0, 1, 2, 3,
et c.
By compar ing t he arguments of the cosines with Eq.
(24-16), it is seen t hat t hey are t he electrica l angles locating

t (Jt)


F ig . 24-7 Exp an sion of the 111111f du o to tho basi c coil sot in t o a Fourie r ser ies (t = 0, i a = V 2I a )

the position of the point x multiplied by v

(vx h ) :rt = va = a v
In other word s, the arguments of the cosines are equal t o
the electrical angles a v for the vth h ar monic with a period
taken equal t o 2:rt:
(vxh) :rt = (xh v ) :rt = a v
Subject to t he qu alifica tion s made as rega rds t he mmf,
t he coefficients, or the amplitudes, of t he var ious harmonic s are given by

+, / 2
F c vm

= 2T


Ft=o cos (vxh) :rt dx = _ 4_ F e m k l'v


- , /2


F t=o = F e m
F t=o = 0


- Ye/2 < x < Yc/2

0:12 > I x I > Yc/2


Part Two. En ergy Conversion by Electrical Machines


Equation (24-18) contains what is known as the pitch

factor, k p , defined by
k p 'V = sin (vY c'It/2T:) = sin (va y/2)
It characterizes the effect ' that the coil pitch Yc and the
chording angle a y have on the peak value of a harmonic
mmf .
For the fund ament al, t hat is , for v = 1,

With a full pitch, that is, when Yc = T:, the pitch factor for
the fundamental is equal t o unity
kp1 = kp = 1

For the higher harmonics (v

kp 'V =


1), it may take values

sin (v'It/2) = + 1


The sign of the pitch factor gives the sign of the harmonic
mmf at the coil axis where x = 0 (in Fig. 24-7, k p 1 and
F c 1m ar e positive , whereas k p 3 and F c 3m are negative) .


Ye = ~ t


Fe l m



Fig . 24-8 The effect of pitch-shortening (chording) on the mml harmonics

In the li ght of t he for egoing, the pitch factor may be

construed as the ratio of the peak value of a harmonic mmf
in a given coil to the peak value of t he same harmonic in
the case of a full coil pitch , that is, when Yc = T:.
It. foll ows from E C[ . (24-19) that the effect of the pi tch
factor on the mm f vari es with th e elect rical angle spanned
by the coil and the ord er (or number) of the harmonic.
This effect is a maximum for the fundam ental whose peak

Gh. 24 Mutual Magnetic Fi eld of Phase Winding


value is
F c 1m

= 4Fc mk p 1/n


The higher harmonic mmfs have substantially lower peak

values (their absolute values are meant)

I r-; III r-: I = I k p v Ilv I k p 1 I

If we choose the coil pitch such that

Yc = (v - 1) xl
the vth harmonic mmf will be nonexistent. This can be
proved from Eq. (24-19) on recalling that v is an odd number
and, as a consequence, v - 1
is always an even number.
With the coil pitch thus
chosen, the pitch factor for the
vth harmonic will be zero
. '
nv = SIn 'V'L
: =sinkn=O
where k is an integer.
As an example, when



the third harmonic mmf will,

as is seen in Fig. 24-8, be
nonexistent (F c 3 m = If,P3 = 0).
. For a better performance of
Fig. 24-9 Plots of kpv as a
the machine, it is desired that
function of Yc
the mmf should be sinusoidally or cosinusoidally distributed in space. Therefore, the pitch factor should preferably
be chosen such that the higher harmonic mmfs are minimized.
This cannot, however, be done for all the higher harmonic
mmfs at the same time, because in order t o eliminate any
particular harmonic, the pitch factor must have a particular
value. The best that can be done is to strike a balance by
choosing Yc ranging between 0.82'L and 0.85'L. Then, as is
seen from Fig. 24-9, the fundamental mmf will remain
about the same as with a full pitch (k p 1 = 0.96 to 0.98),
whereas the fifth and seventh harmonics will be substanti-0.6



Par t Two. En ergy Conversion by Ele ctri cal Machin es

ally att enua t ed (k p 5 = 0.16 to 0.35, and k p 7 = 0.35 t o 0.08).

Unfortunately, t he third harmonic still retains a mark ed
value, but it can be eliminated from the resultant mmf by
other means (see Sec. 25-4). As reg ards the st ill hi gher
harmonics (the 11th, 13th, 15th, etc .), they ar e substa nt ially lower in peak value than the fu nd am ental. E ven wit h
high values of the pitch factor,
c1ll 7
F e l m -- I.F
F CVm _4F
- - - /ipv ~
1 e m ,e P I :rt

24- 5


The Phase MMF.

The Distribution Factor

With an arbitrary number q of coils per group , a phase of

a winding ma y be im agined consisting of q basic coil sets
(elementary phases) in which there is only one coil per pole
pitch (see Sec. 24-3). Such a basic coil set (an elementary
pha se) has q = 1. For example , phase A in the winding of
Fig. 22-3 consists of q = 3 basic coil sets (elem entary phases),
namely: t he basic set of coils 1, 10 , 19, and 28 (shown by
heavy lines), the basic set of coils 2, 11 , 20, and 29, and th e
basi c set of coils 3, 12, 21, an d 30. Therefore , with an arbitra ry q, the ph ase mmf can be found as the sum of the mmfs
due t o t he var ious ba sic coil sets.
. For this sum to be taken analytically, it is convenient
first t o find the various harmonic compo nents of the ph ase
mmf as the sum s of t he respect iv e harmonic compo nents of
mmfs due to the basic coil sets . Let us take this sum, beginning with the fun damental com ponent, t hen for the hi gher
order harmonic components of the phase mmf on the assumpt ion , as before, that the phase curr ent is a maximum
ia =


Th e fundam ental mmfs for q basic coil sets with peak

values F cI m are sh own in Fi g. 24.-10 . Th e b asic coil set
labelled "1" is made up of the firs t coil s within each pole
pitch. The basic coil set numbered "2" consists of the second
coil s, and so on. Neighbouring basi c coil sets are displaced
from one an other by a tooth . (or slot) pitch t z along the
periphery of the air gap , and t he fu ndam ental mm fs in the
elementary phases are displaced from one another by the
electrical slot (or t oot h) angle a z = t z:rth: = (2:rttZ) /p .

Gh. 24 Mutual Magn etic Fi eld of Phase Winding


To sim pli fy the matters, the figure shows full-pitch (unchorded) coils. In t aking the sum of the fundam ental components
of mmf set up by the basic coil sets, it should be rem emb ered
that they are displaced from each other by the tooth
angle a z - and their axes 1 , 2, 3, and 4 passed through
Phase a xis

1 2 s Ii(h) (g)



Fig . 24-10 The funda mental mm fs due to t he basic coil sels making
up a ph ase with q = 4

the peaks of the cosinusoidally distributed mmf are displa ced by the angles a 0 1 , a 0 2 ' 0.:0 3 = aO It and ao~ = a oq
from the phase axis (the latter being the axis of symmetry
of the coil group within a given pole pitch) .
The electrical angle between the axis of the nth basic
coil set and the phase axis is
that is,

a OIt

= az (n - 1) - a z (q - 1)/2
a Ol

a 02

= - 3a z /2
= -o.: z/2

0.:0 3

a o~



a z /2
3a z/2

Th e pha se mmf equal to t he sum of the mmfs due to

the basic coil sets is cosinusoidally distributed over a pole
F = F ph 1m cos a
The phase mmf has a peak value, F phlm , at the phase axis


Part Two. Energy Conversion by Electrical Machin es

and can he found as the sum of the mmfs produced hy the

various basic coil sets at the phase axis:

F phlm = ~ F elm cos CX On



In developing an analytical expression for the peak value

of the phase mmf, it is convenient to write the mmfs due
to the basic coil sets as complex amplitudes

Fnm =

F e 1 m exp (jcxon)
shown for n = 1, 2, 3, 4, ... , q in Fig. 24-10. If we align
the phase axis with the real axis of the complex plane, the
mmf due to the nth basic coil set at the phase axis will he
equal to the real part of the complex amplitude, F nm :
F e 1 m cos

CX on


= Re {F nm}


exp (jcx on) = cos CX on

j sin CX On
Hence the peak value of the phase mmf will he

F p h 1m= ~ Re{Fnm}=Re

r 2J


Fnml =Re{F p h 1m }

As is seen, the complex amplitude of the phase mmf



is the phasor sum of the component mmfs due to the

basic coil sets. In Fig. 24-11, their sum is taken on a
reduced scale.
Noting that the polygon formed hy the complex amplitudes of mmfs being summed can he inscribed in a circle
of radius
F e 1m/2 sin (cxz/2)
we can find the peak value of the phase mmf, F p h i m- from
the right-angled triangle ODA

F p h 1m = 2 (OA) sin (qcxz/2) = qF elmkdl = 2

yex Iwk o1/np

where I = ala = rms phase ;~current
w = 2pweq/a = turns per phase path
kW 1 = kp1k d 1
= winding factor for -the fundamental component of mmf.


Ch. 24 Mutual Magn etic Field of Phase Winding

E quation (24-28) contains wh at is known as the di stribution

factor for the fundamental component of phase mmf:
k d1

sin (qaz/2)/q sin (az/2)


It is t he ratio of the peak value of the fundamental mmf of

a phase to the arithmet ic sum of the peak fundamental mmfs
Phase axi s

Phase asis

o -)

0 000
f Z J If


Fig . 24-11 Combi nin g th e mmfs due to the basic coil sets in Fig. 24-10,
ma king up a phase

clue t o the basic coil sets in that phase

k d1

F P lil m/ q F c 1m

In finding the peak value of the vth harmonic component

of the phase mmf, it sh oul d be rememb ered t ha t the respecti ve angles, a z v , are v t imes as great as for t he funclamen tal:
Th erefore, the peak value of the v th ha rmonic of the phase
mmf will be

si II (i'qa z /2)

k cl\, = qsin (v IXz/2)

is known as t he di stri bution factor for the vt h harm oni c.

Part Two. En ergy Conversion by Electrical Machines


In an m-phase symmetrical winding, a coil group spans

1/mth fraction of a pole pitch or the electrica l angle sclm.
= qaz determined for the fund amental. Therefore ,
k = s in (v n /2m)

q si n (v n /2mq)

Thus, for a three-phase winding, where m. = 3,

k d v = sin (vJt/6)/q sin (vJt/6q)
Using Eq s. (24-31) and (24-33) , it is an easy ma tter t o
trace how an inc rease in q, the number of coils per group,





IJ =1






Fig . 24-12 Diagrams of

in g)

k dv

as a fun cti on of q (for a t hree-phase wind-

can affect the phase mmf waveform . Fi gur e 24-12 gives the
distribution factors for the fund amental and higher-order
harmonic components of phase mmf for several values of q.
As is seen, when q = 1, all distribution factors are unity.
As q is increased , k d 1 decreases insignificantly (when q = 2,
it is 0.969, and when q = 00 , k d 1 = 0.9 55). In contrast ,
the distribution factors for t he higher harmonics go down
abruptly as q is increased, so that wh en q = 00 , the distribution factors for the triplen harmonics becom e
I k d v I = 2k d / v

6h. 24 ~iutual Magnetic Field of Phase Winding


and for all the other harmonics,

I kd'V I =


As is seen, even with moderate values of q (say, 3 or 4),

the distribution factors are about the same as they are
for q = 00. The only exception is the so-called slot harmonics
(slot ripple) whose order (number) is given by
v = kZ/p + 1 = 2mqk + 1
where k is any integer. For them, the distribution factor is
equal to that for the fundamental component, kd'V = k d l *.
For example, when q = 2, this property is manifested by
the harmonics of order v = 2mqk + 1 = 2 X 3 X 2k
+ 1 = 11, 13, 23, and 25. This can readily be verified
by reference to Fig. 24-12 where the distribution factors for
slot harmonics are shown shaded.
As q goes up, the higher harmonic components contribute
progressively less to the phase mmf (the only exception
being the slot harmonics). Importantly, in the phase mmf
their effect is less noticeable than in the mmf due to a basic
coil set [see Eqs. (24-29) and (24-31)1,

In the limit, for a uniformly distributed winding (with v

other than a multiple of three, and also for other than slot
Because in the phase mmf the slut harmonics are present
to the same extent as in the mmf due to the basic coil set
[see Eqs. (24-26) and (24-31)],

To minimize their effect, it will be a good plan to avoid

t he values of q that are less than three. However, already
at q = 3 the order of slot harmonics,


+ 1= 2

X 3 X 3

+ 1 = 17 or 19

* For slot harmonics (slot ripple), the pitch factor, too, is the
same as for the fundamental, that is 1cp 'V = k p i-


Part Two. Energy Conversion by Electrical Machin es

is so high that even with k d v = k d 1 and k p v = k p 1 thes e

harmonics are only slightly present in the phase mmf
F ph17m

F phlm/17

F ph19m = F Phlm/19
With an appropriately chosen va lue of Ye and a sufficiently
large number of coils per pole pel' phase , the phase mmf can
be made sinusoidal very nearly. When the degree of chording
(short-pitching) is taken equal to its recommended value,
yeh: ~ 0.8, the phase mmf may contain a fairly noticeable
third harmonic . This is, however, of minor importance
Phase axis

Fig. 24-13 The pha se mmf of a three-ph ase winding (m = 3, q = 4,


0.83 5)

because the resultant mmf of a three-phase winding contains

no third harmonic. Figure 24-13 shows the phase mmf and
its harmonics for a three-phase winding with q = 4 and
yeh: = 0.835.
Using Eq. (211-31), the peak value of the phase mmf and
of its harmonics may be expressed in terms of the peak value
of the coil-side current , 11 "21aWe, that is ,

F p h 11l = q 11 "2 I aWe



CII . 24 Mutual Magnetic Field of Phase Winding

The peak values of the mmfs found for the condition s specified in Fig. 24-13 are as follows:
F ph m

4"V2" I aWe

F phlm = 4.741/ 2" Iaw e

F ph 3m = - 0.80S 1/2 t,;

F ph 5m = +0.OS1/2 I a we
F ph 7m = -0 .0281/2 I aw e

Pulsating Harmonics of the Phase MMF

In the previous section, we have seen how the phase mmf is

distributed in space at t ime t = 0, when the ph ase current
is a maximum, i = V"2 I . Because the phase current varies
Phase axis

"1/ 2" I

cos rot

it is clear that at any other

point x (a) in the air gap the

phase mmf will be proportional to the inst ant aneous phase

current . Obviously the spa t ia l distribution pat t ern of the
phase mmf will be the same
as at t = a (see ab ove) . The
solid line in Fig . 24-14 shows
FIg . 24-14 Ripp le in [the funthe fundamental component
damental component of the
of the phase mmf at time t =
phas e mmf
= O. For t ime t l , it is shown by
a dashed line. Spatial variations in the fundamental component of the phase mmf can be described by the following

F (a,t) = F (0, t) cos a = F phlm cos rot cos a


Here, a = xsd-: [see Eq . (24-16)] is the electrical angle defining the position of a given point relative to the phase
axis, F (0, t) is the mmf on the phase axis at a = a and
at time t :
F (0, t) = F phlm cos rot


Part Two. Energy Conversion by Electrical Machines

Equation (2/1-36) is the equation of a pulsating wave;

it enables us to de termine the fundamental cornponen t of
the mmf at any point along the ail' gap and at any time .
For the vth harmonic of the mmf, this equation is written
F (a, t) = F phvm cos wl cos a v
where a v = xnIr: v ,
The axis of the pulsating mmf remains stationary in space
and coincides with the phase axis (see Fig. 25-3).


The Mutual Magnetic Field

of a Polyphase Winding


Presentation of the Pulsating Harmonics

of the Phase MMF as the Sum
of Rotating MMFs

The mutual magnetic field of an In-phase winding is produced by the sum of the phase mmfs . The pulsating harmonics of the phase mmfs can be presented as the sum of revolving mmf waves .
If we write the product of cosines in Eq . (24-36) as the
sum of cosines, we get
F (a, t) = 1/ 2F ph1m cos (wt - a)
1/ 2 F p h 1m COS (wt
= F~I11m cos (wt - a) + F~h 1m COS (wt
= F' (a, t) Fit (a, t)

The first term in Eq. (25-1) is a forward revolving mmf

wave , and the second term is a backward revolving mml
wave. The revolving mmf waves are written with reference
to the phase axis which is assumed to be stationary in space.
To get insight into the basic properties of these waves, let
lis re-write Eq . (25-1) in a rotating system of coordinates.
The state of the forward rotating mmf wave relative to i ts
axis, which also rotates at angular velocity wand coincides
at t ime t = 0 with the phase axis (Fig. 25-'1a), is defined
by the angle au = a - wt, and its equation m ay be written
F' (a, t) = F~ hl m cos (-au) = F~ 111m cos au

Ch. 25 Mutual Magn etic Field of Polyphase Winding


At time t, the forward rotating mmf Wave is shown in

Fig . 25-1a. From Eq. (25-2) it follows that the forward
rotating mmf is a maximum at a o = 0, th at is

F' (a , t)

F~h lm

It will remain unchanged at any point displaced by an

angle a o from the mmfaxis. In other words, the forward
rotating mmf wave remains sta tionary relative t o the mmf
Phase axis

Fig. 25-1 Forw ard rotating componen t of the phas e mmf in (a) th e
model of a machine and (b) in th e machine its elf for p = 2

axis and rotates together with this axis at an angular velocity CD in the positive direction (which is counterclockwise).
At t = 0, the positive m aximum of the mmf wave occurs
at the ph ase axis, CDt = 0.
Figure 25-1b shows a rotating mmf wave in a four-pole
machine. Therefore, all the angles are halved, that is,
reduced, by a factor of p, and the angular velocity of the
mmf is .

= CDlp


Or, in words, the angular velocity of the mmf is 1lp of its

electrical angular velocity which is equal to the angular
frequency of the phase cur rent .

Par t Two. Ene rgy Conversion

Phase axil:
(w t - a) __--1--_


Fig. 25-2 Backward rotating mmf

in the model of a machine

Phase axis


Fig: 25-3 Pulsat in g harmonic of

the . phase mmf as the sum ' of th e
forward and backward ro tating
uuuf wa ves


Eiectrica i MachInes

The rotating mmf completes one cycle of change

in the time span equal to
the circumferential period
of the winding)
21: = 2nR/p
The circumferential linear
velocit y of the forward rotating mmf wave is
u = u~ = QR = 211:
where 1 is the frequency of
the cur rent .
The electrical angle is
given by


= (xh)


where x (y) is the distance

from the phase -axis.
Reasoning as" abo ve and
writing an equation for the
backward mmf, F" (cc, t)
rotating about its axis
which in turn rot at es in
t he
(which is clockwise) at a
velocity (0 and is displaced from the phase axis
by an angle ust, it can be
shown that F" (cc, t) is a
backward mmf wave which
has properties similar to
those of the forward mmf
wave (Fig. 25-2). To be
more specific, both waves
the same
F~ h 1m = F~h


F ph 1m/ 2

and rotate at angular velocities Q~ = (0/p and Q~ = - (0/p,

Ch, 25 Mutual Magnetic Field of Polyphase Winding


Their electrical angular velocities are likewise the same in

magnitude and are equal to the angular frequency of the
phase current
coi = Qip = co
co; = Qip = -co
At t ime t = 0, both waves are coincident in space with
the phase axis (Fig. 25-3a). From that instant on, the forward
wave travels in the positive direction, and the backward
wave in the negative direction. Figure 25-3b and c shows
the positions of the two waves for cot = n/6 and cot = n12,

Presentation of Phase MMF Harmonics

as Complex Time- Space Functions

The mmf at point a and time t, that is, 1'. (a , t), may be
treated as the real part of the sum of some complex timespace functions

F' (a, t) + F" (a, t)

= Re [F~hlm exp (jcot) exp (-ja)]
+ Re [F~hlm exp (-jcot) exp (-ja)]
The complex time-space function

F (a, t)



F~hlm exp (jcot)

describes the forward wave of the phase mmf. The complex

time-space function

Fp h1 =

F p h1m

exp (-jcot)

describes the backward wave of the phase mmf. Therefore,

Eq. (25-5) may alternatively be written as
F (ct, t) = F' (a, t)
F" (e, t)

Re [F~hlm exp (-ja)]


LFp hl m exp


If we plot the complex functions F~hlm and Fp h1m on the

space-time plane of a ' two-pole mod el , in which the real
axis runs along the phase axis in Fig. 25-4, and the imaginary
axis is turned through n/2 counterclockwise, we shall see
that the angle between the point at the angle ct and F~hlm
is (cot - a). Likewise, the angle between the point at the

Part Two. Energy Conversion by Electrical Machin es

angle a and F~Jh1m is (-rot - a). Therefore, as stems from

Eq. (25-1), a projection of F~h 1m or
1m on a direction
at the ,angle a will, respectively, give the forward mmf,
F' (a, t), or the backward mmf, F" (a, t), at the point in
The complex function F~h 1m rotates in the forward direction (in the direction of positive angles) at angular velocity


Phase axis


_ - - Fphlm



I --- {2[ COSCtJt

Fig. 25-4 Representation of the phas e mmf on the complex pl ane of

a two-pole model
oi, whereas the complex function F~h 1m does so in the backward direction at the same angular velocity co. This form
of presentation applies when the angle a is reckoned from
the phase axis aligned with the real axis of the complex
plane, and time is counted from t = 0 when the phase current is a maximum, i =V2I. In the final analysis, however, we are interested in the mmf of a polyphase winding,
and it can be found by adding together the mmfs of the
individual phases.
To tackle this problem, we should learn to write the equation of the mmf for an arbitrary phase whose axis makes an
angle 'aph with the real axis of the complex plane (Fig. 25-5),
and whose current is given by


V2 I

cos (rot- (PPh)

so that at t = 0, the current is


V2" cos ( -



Ch. 25 Mutual Magnetic Field of Polyphase Winding

The equation for an arbitrary phase mmf can be written

in trigonometric or complex form by analogy with Eq. (25-1)
or Eq. (25-5), noting that the angle cat is now replaced by
Phase axis

Fig. 25-5 Representation of an arbitrary phase mmf with an arbitrary ph ase current on the complex plane of a two-pole model

(wt - (Pph), and the angle a by (a - aph) reckoned from

the phase axis,

F (a, t)

F~h 1m COS


[(wt - (Pph) - (a - aph)]

F~h 1m COS [-(wt - (Pph) - (a - aph)]

Re {F~h1m exp [j (wt - (PPh)] exp (japh)

X exp (-jan
Re {F~ll1m exp [-j (wt - CPPh)] exp (japh)
X exp (-jan
F (a, t) = Re LF~ 111m exp (-ja)]

F (a, t)

= Re [F~hlm exp (-ja)1

+ Re

LF~ll1m exp (-ja)]

A plot of an arbitrary phase mmi on the complex plane
of the model is shown in Fig. 25-5. The complex function



+ F~ll1m

describes the phase mmf, the complex function


, .

F~ll1m exp [j (wt - (P Ph) exp (japh)]


, :



Part Two. Energy Conversion by Electrical Machines


describes the forward rotating wave of the phase mmf, and

the complex function

describes the backward rotating wave of the phase mmf.


Time and Space-Time Complex Quantities *

and Functions of the Quantities Involved
in Operation of a Polyphase Machine

Scalars sinusoidally varying in time (currents, voltages,

emfs, and flux linkages) are customarily represented as
complex functions whose projections on the time axis give

wt- 25r

Axis A (t)


Fig . 25-6 Representation of currents in a three-phase mac hine on

the time complex plane (on the left) and on the spac e complex plane
(on the right) of a two-pole model

instantaneous va lues of those quantities. For example,

on the left of Fig . 25-6, the inst ant aneous value of the
phase A current, reduced by a factor of y2, is equal to the
project ion of the complex harmonic function of the phase A
current, fA = I A exp (jwt), on the time (t) ax is aligned
with the rea l ax is of the time comp lex plane or, to state
this differently, to the rea l part of the complex current

* Time varying quantities are usually called pha sors . Spatially

distributed quantities are true vectors. Frequently, they are plotted
to~eLhel' on combined phasor-voctor dia gram .i-- Translator's Ilo ! ~.

Ch. 25 Mutual Magnetic Field of Polyphase Winding


i A /V "2 = Re (TAl = I A cos rot

where I A

rms current in phase A in coniplex notation

I A = rms current in phase A
Under balanced conditions, the quantities in the other
phases of a three-phase machine can be described by complex functions displaced by -2n/3 (for phase B) and -4n/3
(for phase C) from those associated with phase A. For
example, the currents in phases Band C are written in
complex-function notation as

I A exp (jO)


n exp




exp (jrot)

where j n = In exp (-j2n/3) and j e = Ie exp (-j4n/3)

are the complex rms currents in phase Band C, respectively.
Since the rms phase currents are the same, the magnitudes
of the complex currents are likewise the same

The instantaneous phase currents can be found from the
following equations:
i A /l/ "2 = Re

rf,d =

Re [j exp (jrot)J

i n/y


Re {j exp [j (rot- 2n/3)]}

i e /V2 =



Re [Iel = Re {j exp [j (rot- 4n/3)]}

where j = j A' and are each a projection of the respective

complex current function on the real axis of the time complex plane (Fig. 25-6).
The theory of electrical machines uses another form of
representation for the quantities existing in polyphase
systems under balanced conditions. More specifically, scalars
(currents, voltages, and so on) associated with the various
phases are depicted on the space complex plane of a twopole model as a complex function common to all the phases.
For the phase currents defined by Eq. (25-7) and shown
on the left of Fi g. 25-G , such a complex function has the


Part Two. En ergy Conversion by Electrical Machines


T= j

exp (jwt)

wh ere J = J A' an d is pl ot t ed on the space com plex plane

of a two -pole model as shown on the right of Fig. 25-6.
In a two-pole model , t he phase windings are shown each
as a coil traversed by a positive current. The axes of ph ases
A, B, and C are drawn through the centres of the coil groups
represented by a single coil. Because the event s in phase B
lag behind those in phase A, the axis of ph ase B is di splaced
from that of phase A by an electrical angle equal to 2n /3
in the positive direction (counterclockwise), and that of
phase C by an angle equa l to 4n/3 in t he same direction .
The instantaneous phase curre nt (reduced by a fac tor of }/ 2)
is given by a pr ojection of the complex function I on t he
respective phase axis.
Because the complex current fu nction I takes up t he same
position relative to the axis of a given phase as the complex
current function of the same phase re lative to the real ax is
of the time com plex plane (on the left of Fig. 25-6), ei ther
form of representation gives t he sam e inst an t an eous phase
current . To demonstrate, proj ections of t he com plex current
function on the respective phase axes on the sp ace complex

i A /V'2 = Re [/]


'2 = Re

Re [j exp (jwt)]

exp ( - j2n/3]

= Re {I exp [j (wt - 2n /3)J)


ic/V 2" = Re [/ exp ( - j4 n /3)] = Re {j exp [j (wt - 4n/3)J)


are the same as proj ections of th e complex functi ons

and I e on the time axis (see Eq . (25-7) and the plot on
the left of Fig . 25-6). Similarly , we can depict on t he sp ace
complex pl an e of t he model t he emfs , voltages and flux
linkages asso ciated with t he va rious phases . These quantities,
too, will be represented by t he re spective compl ex functions
common to all th e ph ases .
E arlier, we d isc ussed the represen ta tion of spatially di stributed, time-varyin g sca l ars in the form of space-time
complex functions
the space complex
pl ane of a model.

1 B,

Ch. 25 Mutua l Magnetic Field of Polyphase Winding


We did this for the rotating mmf wave which is a scalar

quantity sinusoidally varying with time and space. The
va lue of the mmf at a given point in the air gap, say,
F' (a , t), displaced by an angle a from the origin , was found
for each instant of time as a projection of the rotating complex function F~ll1m on the direction at the ang le a (see
Figs . 25-4 and 25-5). .
Now we have depicted phase sca lar quantities as complex
functions on the same space complex plane . In contrast to
the comp lex functions depicting spatially distributed sca lars
(mmfs, and , as we shall see later , the normal component
of the air-gap magnetic flux dens ity), h owever , t he complex
functions representing phase quantities can only be projected on the phase axes . Their projections on an arbitrary
reference direction have no physical meaning.
To stress this difference, the complex functions of spatially distribu ted, time-varying quantities (mmfs and the
normal component of the airgap magnetic flux density) will
be called time-space complex functions . The comp lex func tions of the phase quantities which only vary with time
(currents, voltages, emfs, and flux li nk ages) will be referred
to as t ime complex functions.

The MMF of a Polyphase Winding.

Its Rotating Harmonics

Consider a symmetrical m-phase winding . To simplify the

matter, let m. be equal to 3. We set out to find the mmf
of this three-phase winding as the sum of the mmfs in the
individual phases. In doing so, we shall remember that ' the
phase axes are disp laced from one another in space by an
electrical angle 2n/m = 2n /3, and that the phase currents
are disp laced from one another by the same ang le in time.
Suppose that the phases carry a balanced set of PPS
currents. Such currents are shown in Fig. 25-6 and can be
found by Eq. (25-7) or (25-8) . As will be reca lled, in a balanced set of PPS currents, the phase B current lags behind
the phase A current by 2n /3, and the index "B" is assigned
to the phase whose axis is displaced from that of ax is A
by an electrica l ang le a A B = 2n/3 in the positive direction,
(see Fig. 25-7).
To combine the ph ase mmfs in complex form, the ax is
of phase A must be aligned with the real axis of the complex


Part Two. Energy Conversion by Electrical Machines

pl ane, and the positive angles cx. must be counted counterclockwise . Resolving the phase mmfs into the forward and
backward components and noting the phase shift f[Jph between the currents and the spatial shift cx.ph between the

Axis A



Fig . 25-7 Produc tion of a rot at ing mrn f by a three-phase winding

carrying PPS currents

phase axes from Eq . (25-6), we obtain the total forward

mmf for a t hree-phase winding as

r.; =

Fim =


+ FBIm+ Feim

= F~ll1m


exp [j (rot - 0)] exp (jO)

F~ll1m exp [j (rot - 2n/3) ] exp (j2n/3)
F~ll1m exp [j (rot - 4n/3)I exp (j4n/3)

= 3F~hIm = 3F~hIm exp (jrot)


Fi m

exp (jrot)

As is seen, all phase mmfs are identical and are depicted

graphically by the same complex function . The peak mmf
of an m-phase winding is

Fim = Fim = mF~hlm = mF p h i n/ 2

where F p In m is the peak va lu e of the pulsating phase mmf
wave . .
Upon suitable substitution s [see Eq. (24-28)1, we get

Flm= Fl m =

(rn. 11

2/n) (Iwk p1k ,lj/p)


Ch. 25 Mutual Magnetic Field of Polyphase Winding


The backward phase mmf waves sum to zero [see Fig . 25-7
and Eq . (25-6)1


F;m = FAl m

+ F1JI m+ FeInt



exp [- j (rot - 0)1 exp (j0)

[-j (rot - 2:rt/3)1 exp (j2:rt/3)
[- j (rot - 4:rt/3)l exp (j4:rt/3)

+ F~ll1m exp
+ F~hlnt exp

= 0
To sum up , the fundamental mmf of a three-phase (or,
generally, a polyphase) winding carrying a set of PPS currents is the forward rotating mmf with the peak value given
by Eq . (25-9). It rotates at an electrical angular velocity ro
and a mechanical angular velocity Q l = ro/p in the positive
direction (counter-clockwise).
On the space complex plane, this mmf runs in the same
direction as the complex functi on 1 depicting the PPS
phase currents (see Fig. 25-7). Recalling that t his mmf is
proportional to current , F l m = kF I, we may re-write Eq,
(25-9) in complex form as



= kF1

= m V2 wkt/:rtp

The distribution of the fundamental mmf set up by the

PPS currents along the periphery of the air gap [see Eqs.
(25-'1) and (25-5)] can be descr ibed by an equation of the
F (ex, t) = F l m cos (rot - ex) = Re [J,\m exp (-jex)] (25-'10)
ex = yp = xsd -t = the electrical angle defining the position
of a given point along the periphery of the air gap
y = the mechanical angle from the origin (from the axis
of the main phase A carrying a current i A = V 2/
X cos rot) to the point in question
x = the distance along the periphery of the air gap from
the axis of the main phase to the point of int erest
L = the pole pitch for the fun damental component
p = the number of pole pairs along the peri phery of air
gap for the fundamenta l component


Part Two. Energy Conversion by Electrical Machines

If a polyphase winding carries NPS currents, the backward phase mmfs will be represented by the same complex
function, and the forward mmfs will cancel out . This results
in a backward rotating mmf which can be described by an
equation of the form
F (a, t) = F 1 m cos (-wt - a)
= Re [F 1 m exp (-jwt) exp (-ja)]

Acting in a similar way, we can combine th e higher

harmonic components of the forward and backward phase
mmfs . In combining, either the forward or the backward
vth harmonic waves, or both, may cancel out . Because of
this, the resultant mmf may only contain some of the harmonics whose order is given by

= 2mc + 1

where c = 0, 1, 2, 3, .. . For a three-phase winding, the

order of the resultant rotating mmf waves will he v = 1, 5,
7, 11, 13, etc.
The vth harmonic component of the resultant rotating
field will rotate in the forward direction (clockwise), if
sign is adopted in finding the order of the harmonic
by the above equation, and in the backward direction, if
the "-" sign is adopted .
The peak value of the rotating vth harmonic mmf can
be found by an equation similar to Eq. (25-11)


m l(Z Iwkdvkpv

vm -


where vp = Pv is the number of pole pairs for the vth harmonic component.
The mechanical angular velocity of the vth harmonic mmf
is given by
Q" = lp ; = lvp
The angular velocity


= w


defined as the product of the mechanical speed by the number

of pole pairs for the vth harmonic field component (which
is also true of the electrical angle for the vth harmonic,
a ; = ,\,p,,) may be looked upon as the electrical angular velocity of the vth harmonic mmf . It is the velocity at which

Ch. 25 Mutual Magnetic Field of Polyphase Wind ing


the complex function of the vth harmonic mmf rotates [see

Eq. (25-10)]
Fv m = F v m exp (+j(j)t)
The mechanical angular velocity of the vth harmonic
is tlv of that of the fundamental component of the mmf,


The distribution of the vth harmonic mmf along the periphery of the airgap is descr ibed by an equation set up by
analogy with Eq. (25-10) (if the harmonic is rotating in the
forward direction) or Eq . (25-11) (if the harmon ic is rotating
in the backward direction):
F v (cc , t) = F vrn cos (+(j)t - Clv)
= Re [F vm exp (+j(j)t) exp ( - j ClvH


sign applies when the harmonics are rotating

where the"
in the positive (forward) direction, and the "-" sign, when
the harmonics are rotating in the negative (backward)

= Pv"? = vp,,? = (xh v ) n = vastl :

is the electrical angle defining the position of a given point

in the fie ld set up by the vth harmonic component
Pv = pv

is the number of pole pairs along the periphery of the airgap

for the vth harmonic component, and

"v =


is the pole pitch for the vth harmonic component.

As a ru le, the vth harmonic is small in peak value, because
the winding factor kdvk p v is only a few hundredths of unity,
whereas for the fundamental component it is close to unity.
Also, many harmonic components cancel out (for example,
this is true of the triplen harmonics in the case of a threephase winding, that is, those whose order is 3, 9, 15, etc .).
Therefore, with a judicious choice of the coil pitch (Yc =
= 0.83,,) and of the coils per group (q;;;;;;: 2), the mmf of
a polyphase (three-phase) winding will differ but little
from the fundamental mmf, because the higher harmonics
it contains are insignificant in their effect. In fact, it may be
treated as the rotating fundamental wave with a peak value

Pa rt Two. Energy Conversion by Electrical Machines

given by Eq. (25-9) and with a mechanical angular velocity

Q = Q 1 = [p, If the winding carries PPS currents and
the current in phase A is

= V'Z I

cos rot

the peak value of the fundamental mmf at time t will be

displaced by a mechanical angle '\' = iatlp from the axis
of phase A (or by an electrical angle ex. = rot).
The above dis tinctions of the mmf induced in a threephase winding are depicted in Fig. 25-8. The phases of the

2VZlaW c


Fig. 25-8 MMF of a three-pha se winding (q = 4, yclr: = 0.835)

winding shown in the figure do not differ from those in

Fig. 24-13. From a comparison of the mmf in a three-phase
winding with that of anyone phase in the same winding
(see Fig. 24-13), it is readily seen that the waveform of the
mmf is improved appreciably, and it appears sinusoidal
very nearly. The third harmonics, rather pronounced in the
phase mmfs (see Fig. 24-13), cancel one another upon com-

eh. 25

Mutu al Magnetic


of Polyphase Winding


bining. It is to be noted that in Fig. 25-8 the mmf F is found

at t = 0, when the current in phase A is a maximum

= V2 I a cos rot = V2'I a

and the currents in phases Band C are the same

iB =

11 2 I a cos (rot -

ic =

V 2 I a cos (rot -

V 2" I a/2
4n13) = - V 2 I a/2
2n13) = -

As will be recalled, I a = IIa is the current in a parallel

path (circuit) of the winding.
The waveform of the mmf can be plotted as for the mmf
of one phase in Fig. 24-13, if we note that the currents in
phases Band C are flowing in the reverse direction relative
to the positive direction of the phase currents. The peak
value of a harmonic mmf can be expressed.in terms of the
peak values of the coil-side currents, using the equation
derived from Eq. (24-35):
2qmk pv k dv


(l/2I a w e )


By the above equation, it is an easy matter to get

r.; = 7.12 11 2' t,;

r.; = 0.075 V2" Iaw e
F =-0.042 V2i,;


The Fundamental Component of the Magnetic

Flux Density in a Polyphase Winding
(The Rotating Field)

As a rule, the pole pitch for the fundamental mmf of a polyphase winding, 't1 = 't, is many times the tooth (or slot)
pitch of the cores, t ZI and t Z2:
'tIlt ZI

= qlml

't 1It z 2


Therefore, in calculating the field set up by the fundamental mmf,

F (CG, t) = F 1m cos (rot - CG)


Part Two. Energy Conv ersion by Electrical Machines

we are in a position to allow for the effect of saliency on

the average by using the air gap factor

from Eq. (24-10) . Then the mean airgap permeance is the

same as in the homopolar case and is given by


11M 6


By definition (see Sec. 24-3), the fundamental radial com.

ponent of the gap magnetic flux density can be written as
B (cc, t) = floAoF (Ct, t) = B l m cos (wt - )


where B 1 m = ~loAoF1m is the peak value of the fundamental

component of the m agnetic flux density .
In a polyphase machine, the fundamental component of
the airgap magnetic flux densi ty is a rotating wave travelling at the same mechanical an gular velocity and having
the same pole pitch and t he same pole-pitch angle as the
fundamental mmf (Fig . 25-9):




= wlp

= 2nlp

Like the mmf, the magneti c flux density B (Ct, t) can be

depicted on a two-pole model (Fig . 25-10) either as a cosinusoidal rotating wave (F ig. 25-10a) or as a space-t ime
complex function
which is in phase with F1 and I. The projection of iJ1 on
an arbi tra ry direction at an electrical angle Ct = PI' from
the origin in the model (Fig. 25-1 Ob) is equal to the radial
component of t he fundamental magnetic flux density at
a mechanical angl e I' from the origin in the prototype
machine (see Fig. 25-9). The origin is usually taken to be
the axis of the main phase (phase A) in which the current
is a maximum
iA =
I a cos wt
at t = O.


Ch. 25 Mutual Magnetic Field of Polyp hase Winding


Fig . 25 9 Rotating magnetic field in t he ai rgap of a polyp hase

2p-po le m achi ne

(a )

Fig. 25-10 Itoprosont ation of th e rotating magnetic fie ld in th o model

01 Fig. 25-9

20- 0169



Part Two. Energy Conversion by -Electrical Machines

Magnetic Flux Density Harmonics in

the Rotating Magnetic Field
of a Polyphase Wi nding

The magnetic flux density harmonics present in the field

set up by a polyphase winding are functions of both the
spatial distribution of the mmf in . the winding and the
saliency of the cores. Unfortunately, the saliency affects
different harmonics differently, and the air gap fac tor used
in calculating the fundamental component (see Sec. 25-5)
does not permit the higher harmonics to be found wi th
sufficient accuracy.
This is the reason why the first step in determining the
magnetic field set up by a polyphase winding is to construct
a stepped waveform for F, the phase mmf. Prior to that,
we must calculate the instantaneous-phase currents i A , i B ,
and i c, the coil-side currents +t~wc/a, +iBwC/a, and
+icwc/a, and the slot currents is(k-l), is(k)' and ' is(k+l)'
This can readily be done, once the winding circuit is known.
The stepped F waveform shown in Fig. 25-8 has been
constructed within a pole pitch, 'to In Fig. 25-11, a similar
mmf waveform is constructed for a half pole-pit ch , 't/2.
In this case, we assign an arbitrary magnetic potential,
say, qJi, = 0, to one of the tooth (or slot) pitches, say, k ,
Then the potential at the next adjacent tooth pitch, k
1)th slot with a cur rent is(It+l) will be
following the (k



+ is(k+l)

(As will be recalled, the current in a slot is assumed to be

positive when it is flowing outwards, that is, toward the
reader.) Once the potentials in the tooth pitches lying between two adjacent poles, or within two pole pitches, have
been found , t he next step is t o determine the ave rage potential
qJav =

1 ""\:1

II: Li


h= 1

and the mmf, F , of the pol yphase winding for each of t he

t ooth pitches. It is measured over and above the average
potential, qJav, and is equal to
(PIt-l =

qJk-l -



Ch. 25 Mutua l Magnetic Field of Polyphase Windin g

in the (k - 1)th tooth pitch,


'Ph = 'Ph


in the kth tooth pitch, and

(PhH :- (PhH -


in the (k + 1)th tooth pitch, and so on.

The mmf thus found is shown in Fig. 25-1'1. It ma y be
taken equal to t he sum of tooth-pitch mmfs, (Ph (x) . Accord ,"/2



C C1




x /<o


Fig. 25-11 The magnetic field of a polyphase winding and the specified distribution of instanta neous currents among th e phases

ingly, the magnetic field set up by a polyphase winding

may be defined as the sum of the elementary fields estahlished by the tooth-pitch mmfs (Ph (x). The distribution of the
mmf for th e kth tooth pitch is shown separately in


Part: Two, Enr.rgy Conversion bv Elec tr ical Machin es

Fig . 25-12. As is seen, it is rect angular in shape, being

(Ph (x) = lPh inside the tooth pitch, and
(Ph (x) = 0
outside t.he tooth pitch. This implies that in calculating
the field t.he following boundary conditions must. be assum-

Fig . 2fi-t2 The field set up by the mmf of t he kth to oth (slot) pitch

ed : !P = 0 for core 2, !P = !Ph for the kth too th pitch on core 1

1)th and (k - 1)th and all the
and !P = 0 for the (k
other too th pitches on core 1.
The fiel d thus obtained is marked ly affected by the shape
of the slots and air gap. To simplify calculations, onl y t.he
shape of the excited core, 1 , is accurately reproduced,

Ch. 25 Mutual Magnetic Field of Polyphase Win ding


because its slots carry the winding in question . The unexcited core , 2, is replaced by a smooth one, and the effect of
its saliency on the air gap permeance is accounted for
approximately b y introducing an equ ivalent ail' gap

5" = tJk 6 2
[see Eq. (24-10)1.
The scalar magnetic potential <P (x, y) within the fie ld
region can be found analytically, us ing Eq. (23-8) . Then
the magnetic fl ux density is found at t he sur face of t he
smooth core, 2, due to the mmf <Pk (x) :
__ arp (x , y)
B k (X ) -flo

The plot of BI< (x) is shown in Fi g. 25-12. Its shape depends

on two ratios, namely bs/ 5" and t z /5". Because of this, the
waveform of magnetic flux density for any other tooth
pitch , say, Bh+ 1 (x) or B k _ 1 (z), is sim ilar in shape to t hat
of B k (x). Thei r ordinates, however, are multip lied by
<Pk+l/<Pk and (jlk-l/CPk, respectively. Figure 25-11 shows how
the resultant B produced by the mmf of a polyph ase winding
can be found by combining the magnetic flux densities
of the individual tooth pitches, B k (x), Bh+1 (x), and
B k - 1 (x) .
The harmonic components of the resultant magnetic flu x
density may likewise be found by combining the harmonics
of the magnetic fl ux densities in the ind ividu al tooth pitches, B k (x), B h+1 (z), and B k - 1 (x) . Because their respective
waveforms are similar in shape, it will suffice t o apply
Fourier analysis to anyone of them , say, B k (x).
Since the B k (x) waveform is symmetrical about t he centre
of the tooth pitch (see Fig. 25-12) , its Fourier series contains
only a constant component and cosine terms

B k (x ) = B k a v

+ v=1
2.j Bl<vm cos (VXI< 'JT,/T;)

where v = 1, 2, 3, ....
The peak va lue of t he vth harmon ic in the Fourier series
is given by

V ,'fh- (I Xk
B fJ vm = -1 ~ B 11. (X ) cos 't'


Part Two. Energy Conversion by Electrical Machines

Unfortunately, the analytical expression for Bh. (x) is so

elaborate that the integral can only be evaluated numerically on a digital computer. To avoid cumbersome computat ions in engineering applications, Soroker'" has proposed
to express the peak value of the vth harmonic, Bh.vm, in
the Fourier expansion of Bh. (x) in terms of the vth harmonic, B h ov m, taken from the Fourier expansion of an idealized rectangular magnetic flux density Bh.o( x) with a peak

f!oCJJh. I6"

Owing to the symmetry of the B h O (x) waveform about the

centre of the tooth pi tch (Fig. 25-12), its Fourier expansion
likewise contains only a constant and cosine terms


( X0) = B

u. av

LJ B /{Ovm

cos - vx/{lt


,\,=1, 2, 3, .. .
The peak values of the expansion terms can readily be
found analytically


B h Ov m = -:r


B hO ()
x cos -VX/llt
1:- X h =
f!o u



is the peak value of the vth harmonic mmf in the kth tooth
pitch, CJJh (x) .
The vth harmonics of Ell. (x) and B h O (x) are shown in
Fig. 25-12. As is seen, they have the same pole pitches,
= .. Iv, but different peak values, E h v m and B h.ovm' The
ratio of the values of the magnetic flux density harmonics
found with and without allowance for the effect of saliency
is termed the slot factor for the vth harmonic



Soroker T.G., Electrotechnlcky Obzor, 1972, 10.


Ch, 25 .Mutual Magnetic Field of Polyphase Winding

It is the same for all tooth pitches of a given core and solely
depends on its relative dimensions and the harmonic number


t (bsU),',

bsltz, Z/vp)

where Z = number of teeth on the core

p = number of pole pairs for the fundamental component
bs = slo t width at the air gap
From statistical analysis of the numerica l values of C v
found for various relative dimensions, t he following approximate procedure has been proposed for its calculation .
The slot fa ctor for the vth harmonic is

C; = Dv-Av/tan -



where Il ; and A v are found, subject t o th e r atio bs/6" an d

the value of
Cv= -

(i) For e,

txpv bs
- - Z

A v= 1+51l"/
bs ('1 -

S:: "lb s )

' (PmACv U


(Pm A

0.4.845 -

= 1-

an d

_ 05

crmB -

(ii) For



0.0255 bsW'

+ 0.014.2 (bs/6")2

cr mBe~ (1 -

(PmBc U6)

2 (6"lb)2
s -

3 (1 +0.08bs /o")


A v = exp (-1.46 c v6"l bs ) sin (0.95 e v

D v = exp (-1.4.6 c,,6"l bs ) cos (0.95 e v


cr mc)


(Pm C

0 .7484. - 0.05037 bs/6"

+ 0 .001195 (bs/6")2

As an exam pl e, we shall trace the ca lc u la tion of C v for

[he cores in Fig. 25-11 or 25-12 :

8 = '1 mm, bs = bS l = 5 mm, bS 2 = 3 .75 mm


t Z1

10 mm , t Z 2

7.5, mm, Z

24, p


Part Two. Energy Conversion by Electrical Machines

We shall carry out the calculations for the first tooth

(slot) harmonic of order
v = Zip - '1 = 24/'1 - '1 = 23
The airgap factor for the second core [see Eq . (24-'10)) is
k 0 2 = 7.5/(7 .5-'1.607 X '1) = '1.272
The term "'1.607" is given by
'\' = (3.75/'1)2/(5 3.75/'1)

The equivalent air gap is

6" = 6k 02 = 1 X '1.272

'1.272 mm


The values of the other quantities, as Iouud by l.lie equations given above, are as follows:
Cv =

tan (npvIZ)

'1 .5053 < 2

-0 .13'1G5

= 0.60367

A v = 0.43'198

4JmB =

D v = 0.4'1564
C" ~ 3.7'13

Once C; is found, it is an easy matter to determine j he

peak value of the vth magnetic flux density harmonic due
to the mmf of the kth tooth pitch, (Ph (x), with allowance fOI'
the saliency of the core

where (Phvm is the peak value of the vth harmonic mmf over
the kth tooth pitch.
Knowing the spatial distribution of the vth harmonic muif
over the kth tooth pitch

(x) =


cos (vxllnh:)

and using Eq . (25-22), we can readily write an equation for

the distribution of the vth harmonic of the magnetic flux
density with allowance for saliency
B hv (x)

= Bhvmcos


Ch, 25 Mutual Magnetic Fi eld of Pol yphase Winding


So that we coul d take the sum of the i ndivid ual magnetic

flux densities, we must write the above equation in a coordinate system common to all the loops and having its origin
on the axis of phase A (see Figs . 25-1'1 and 25-12):

B k v (x)

B h v m cos (x -

Xh O)


Here , x - X h O = X k is the distance from t he axis of the kth

t oot h to a given point in t he air gap, X h ll is t he dis tan ce from
the axis of phase A to the ax is of the kth tooth, and x is the
distance from the axis of phase A to the point in question.
Noting that xsd-t = a is the electrical angle from the
axis of phase A to the point in qu estion for the fundamental
com ponent, and X h On h = a hO is t he elect rical angle from
the ax is of phase A to t he axis of t he kth tooth , we ma y rewri te t he equat ions for t he spatial disribution of t he v th
harmonics of the mmf and magneti c flu x density as
(P" v (a) = (Phvm cos v (a - a"o)
B h v (a) = B h v m cos v (a - a"o)


Because the currents traversing t he ph ases of the winding

have an angula r fr equen cy (1), we may argue that the mmf
over t he kth too th pitch vari es intime wi th t he same frequency
(P" = (P"m cos (wt - Bk)
where (P'lm is the time peak value of the mmf over the Hh
tooth pitch , an d ~" is t he time phase of the mmf over th e
kth tooth pitch .
The peak va lues of t he space harmon ics of the mmf an d
magn eti c flux density in the kth tooth pi tch will vary in
t ime in t he sa me manner:

B II vm =
where CPhvm


/loC v

~ (Ph vm

(2(p,,/vn) sin (vt zn /2l:) is the peak value of

the vth h armonic mmf at time t, and
(2(Pkm/v n) sin (vt zn /2l:) is th e peak value of
t he vth ha rm onic mm f at th e time when

(P" (t)


Noting that in Eq. (25-23) both CP"vm a nd B "vm are fun ctions of time, we may write the following equations for the
vth harmonics of the mm f and magnetic flux density at any


Part Two . Ene rgy Conversion by Electrical Machines

point in the air gap at an angle a to t he phase axis at any

instant of time t:
(wt :-- ~ 'l ) cos v (a - a/1o) (2.5-24 )
= (/loCvI8")(Pllvmm cos (cut - ~h) cos v (a - a/1o)
On comparing the above equations with Eq .(24-37) , it ca n
be seen that they describe pulsat ing waves.
The resu ltan t vth harmonic of the magnetic flux dens ity in
a polyphase winding is obtained by combining the magnet ic
flux densities due to the mmfs over the tooth pitches :
B ,. (a, t ) = :lj B "v (a , t) = !!6~v 2j CP /1" (a , t)
(a , t)
B/1 v (a, t)

CP1lvmm cos

/1 = 1

/1 = 1

The sum of the vth harmonic mmfs of all t he tooth pitches




(a, t) , is equa l to the vth harmonic mmf of th e


polyphase winding , F v (a, t ).

Earlier (see Sec . 25-4), it has been sh own that if v = 2 me +
+ 1 (where e = O. 1, 2, 3, ... ), an m-phase winding will
generate the vth harmonic mmf as a rotating wave which
can be described by Eq . (25-1!1). Therefore ,



(a , t) = F v (a , t)

= F vm cos (+cut - va)

As a conse quence', the vth harm oni c of t he magnetic flux

density in a polyphase winding is given by
B,. (a , t)

B vm cos (+ wt - vee)


where B v m = ~loC vFvn,l8" is the peak value of th e r ot at ing

wave of the vth harmonic magne t ic fl ux density in a polyphase wind ing .
To sum up, the magnet ic flux dens ity in a polyphase winding , as found with allowance for the effect of slots, contains
the harmonies of the sa me order as that of the mmf harmonics. Th e effect of slot s on the peak value of the magnetic
flux density harmonics is accoun ted for by th e factor C v
calculat ed by Eq . (25-21).
The slot factor may be positive or negat iv e . Accordingly,
the magnetic flu x dens ity wa ve may be in phase or in antiphase with the mmf wa ve. For harmonics with l arge pol e

Ch. 25 Mutual Magne tic Field of Polyp ha se Winding


pitches and sa Lisfy ing the condition Tv = T/V t Z 1 ' C; ~

tlk o! ' Since for the fundamen tal component this condition
is usually satisfied


T/ m 1q1

th e peak va l ue of the associate d mag netic Ilu x densi t y as

given by Eq . (25-25) is


F17Il ~tO Cl/ 6" = Flm ~to/ 8kolk 62

which chec ks with Eq. (25-18) .

The vth harmoni cs of th e magnetic flux density in
the case of a rotating fie ld have the same number of pole
/lxi~' A

AXi s A

S?I t

.. I




I; ig . 25-13 The effect of higher harmonics on th e waveform of th e

rot at ing field set up by a polyphase windi ng:
(a) field at t = 0, (b) field at t = n /2w

pairs , the same pole pitch, the same sense of ro tation , and
the same electrical angular frequenc y as the vth harmonic
mmfs [see Eq . (25-13)J. From the fundamenta l fl ux dens i ty
the y only differ in the much smaller peak values, the num ber of pole pairs , and the me chanical frequen cy of rotation .
Because each harmonic component of the fie ld travels at
its own m ech an ical angular veloc ity Q v , their relative
position is changing all t he Lime, and th e res ultant fie ld
patt ern goes t hro ugh a cycle of change per io dica lly. This
proper ty of the fie ld is ill ustra te d in Fig . 25-13 where the
mag netic flux dens ity is shown as the sum of t he fun damen tal


Part Two. Energy Conversion by Electrical Machin es

and the fifth harmonic . The magnetic flux dens ity is given
for two in st ants , na mely t = 0, whe n the current in phase A
of the t hree-phase winning is a m aximum (see Fig. 25-13a),
and t = n/2w , when the phase A current is zero (see
Fig . 25-13b). At t = 0, t he peak va lu es of the harmon ics
occur on the ph ase ax is . During the t ime t = n /2w the Iunrlament al wave travels in the positive direction th rough a mechanical ang le
Q i t = (wlp ) (n /2w) = n /2p
or an electrical angle n /2, whe reas the 5th harmonic wave
t ravels in the opposite direction through a mec han ica l angle
Q 5 t = (wI5p) (n I2w) = n/2 (5p)
or an electrical angle
(5p) (Q5 t ) = n/2
Referring to the figure, the magnetic flux density waveform
at t =
differs fr om t hat at t = n /2w because the relativ e
positi on of the harmonics is cha ng ing all t he t im e. (For
convenience, the fifth harm oni c is shown enlarged fiv efold.) When the contribution fro m t he h ighe r ha rmonics is
insignificant , th is change in sha pe is negligible .
The pr operties of the h igher harmonics listed above are
typical of the rotating fields produced by polyphase windings
car ry ing ba lanced sets of PPS or NPS curre n ts with a circ ular freq uency ui.
To sum up, t he harmon ic components of the field set up
by a pol yphase winding rotate all at the same electrical
velocity W v = ro which is the same as the ci rcu la r frequen cy of the currents , but with different mechan ical angular
velocities, Q" = talvp .


The Magnetic Field

of a Rotating Field Winding


The Magnetic Field of a Concentrated

Field Winding

Another way of pr oducing a rotating field is to place the

field win ding on the r ot or of a mach in e. When this win ding
is energized with d. c. , it establishes a magne tic fie ld s ta tio-


Ch. ' 2fl Magn eti c Fielrl of Rotating Fip.ld Winding


nary relative to the ro tor, with a radial component of magnetic Ilux den sity B (Fig . 26-'1). If, now, t he ro tor is made
to rotate at me chanical angular velo city Q , the magnetic
field set up by the rotor winding will likewise rotate with
the same angular velocity.
The mmf F produced by a concentrated field winding can
be depicted by a rectangular waveform (see Sec . 24-1) . It
remains constant and equal to Pm = iw c over a pole pitch.
At the slot axis, it changes
r--r-----,r--~~F--.., abruptly by an amount equal
t o the slot current , 2iwc , and
, reverses in polarit y, t urn ing
t o - F m - The peak value of
t he mmf can be found as for
a single-phase, double-layer,
full-pitched winding for which
q = 1 and Yc = T and which
carries a direct current,

= V 2"


F m = q (V

2. fa)



The air gap field set up by

F is cal culated over a half-

pole pitch by the Laplace

equa t ion , (23-8), for a scalar
magnetic potential under t he
Iollowing boundary con ditions: the potential at the surface of the pole-shoe is rp; the
Fig. 26 -1 The magn etic fi eld
potential at the surface of the
of a concen tra ted field winding
smooth core and at the slot
axis is zero . ,
The shape of t he waveform depicting t he radial component
of magnetic flux density at the surface of a smooth core , B,
depends on th e pole en clo sure ex = b"IT, the relative air
gap at the pol e tip y = 8m /8, and the re la ti ve air gap at the
pole axis, e = BIT. The magne tic flux density waveform
shown in Fig. 26-1 has been plo tted for ex = 0.55, Y = 2,
and e = 0.01 .
It is usual to generate magnetic flux density wav eforms
on a computer for various valu es of ex, 1', and f , and to sub ject them to Fourier analysis . The peak values of the various

Part Two. Energy Conversion by Elect rica l Machines






0.7 0.25


0.5 0.15



0.3 0.05

0.1 /).05

F ig . 26-2 The coefficients of

the excitat ion field determinin g the mean mag netic flux
density: af= at
= Bmean/Bm' and its funda mental: kf = kt



0.5 0.6

0.7 0.8

0.9 1.0

F ig . 263 Coefficients of the

excitation field determining the
hig her harmonics of magnetic
flux densi ty : k fV = k fv + k/v

eh. 26 Magnetic Field of Rotating Field Winding


harmonics are then expressed as fractions of the maxim um

flux density, B m , called the harmonic coefficients of the
excitation field:
k f = Blm /E m for the fundamental

kf v

= B vm/ B m for the vth harmonic

(2G .2)

Here, B m = ~toF m/8 is assumed to be the magnetic flux

density set up ill a uniform air gap 6by a constant mmf, Fm*The most accurate values for k f and k j v for v = 1, 3, 5,
7, 9, tl , 13, 15, and 17 can be found in [38]. We will only
give those required to calculate the harmonic coefficient
for the fundamental , k f (Fig . 26-2) and for the 3rd and 5th
harmonics, k f s and k f 5 (Fig. 26-3) . Referring to the figures,
we can find the components of the respective harmonic
coefficients, namely kj and kj, kjs and ki ;j , and kj5 and ki5(As is seen, the figures give yki, yki 3 , yk j5') The harmonic
coefficients are found by combining their components for
the specified va lues of a, y, and B:
kf = kj

+ ki,

kf v = kjv

+ kiv

The B waveform differs in shape from the mmf waveform

and , with a judicious-choice of the relative air-gap (limensions, it can be made sinusoidal very nearly . The magnetic
flux density waveform can be expanded into a Fourier
series where the equation for the vth harmonic about the
winding axis is

B (a) = B v m cos a ov


where a ov = va o = VPYo
a o = electrical ang le defining the position of a given
point relative to the winding axis, and
Yo = mechanical angle defining the position of the
same point relative to the winding axis

The Magnetic Field of a Distributed

Field Winding

The mmf produced by a distributed winding can be depicted

by a stepped wa veform (Fig. 26-4) sim il ar t o that for the
phase mmf of a double-layer winding . For a single-layer
distributed winding with a slot current hUe and with q
* In thi s cas e, the scalar ma gneti c potential is 'P = Fill'


Pnrt Two. Energy Conversion hy Electricnl Mnchi nes

woun d slots per pole, the pea k va lues of the harmon ic mmfs
can he found by Eq . (24-35) der ived for a phase of a doub lelayer wind ing, assuming that the winding is fu ll-pi tched
(Ye = L)* and that the maximum coil current in a doub leWinding axis

F ig . 26- 4 Th e ma gneti c field set u p by a distributed field winding

(q = fl, bIT = 2/3)

layer winding, V 2f aWe ' is equal to half the sl ot current in

the fie ld win ding, tw e .
Not ing that for Ye = r, the harmonic pi t ch fac tor is u ni ty, k p ,. = 1, Eq . (24-35) can be re-written 1.0 give the following expressicn for the peak value of the vth harmoni c mm f:

2qlcd v
----m;--110 c =

4kd v .



10 = we q/2 = turns per pole of the fie l d win ding

k clv = Si ~(q(v'\'z//22 = distrib ut ion factor for the vl.h har-

qSITI V'\'z

l' z =tz;rr,/-r = p;rr, /q = electrical ang le bet ween adja cent wound slots

- - --

* As regard s the gener a t ion or a ma gn et ic field , th e field winding

may be treated as a full-pitched winding, becau se the distance between adjacent groups of wound slots is equal to the po le pitch.

Ch. 25 Magn etic Field of Rotating Field Winding


t z = tooth (slot) pitch

= bl = enclosure of the wound part of a pole

b = length of the wou nd par t of a pole pitch

In this case, the air gap permeanc e may be deemed constant and equal to
over the entire length of the pole pitch. Therefore, t he magnetic flux density waveform , B = f.toFAo' is the same in
shape as the mmf waveform, and the peak values of the har , monic flux densities are proportional to those of the harmonic mm fs
B; = f.toFvAo
The equation for the vth magnetic flux densit y harmonic,
referred to the winding axis, does not differ from that
for a concentrated winding, Eq. (26-3).

The Rotating Harmonics of the

Excitation Field

As the rotor rotates at mechanical angular velocity Q , the

excitation fi eld an d its harmonics (Fig . 26-5 shows onl y the
fundamental and the 5th harmonic) ro tate all at the same
mechanical angular velocity Q . This is the reason why, in
contrast t o the ro ta ting fi eld set up by a polyphase winding,
the field established by the field winding rem ains unchanged
in shape as it rotates.
In contrast, the electrical angular velocities of the various
harmonics are all different

As is seen, it increases with the harmonic. order. (Compare

it with the field set up by a polyphase winding, where t he
electrical angular velocities are the same, but the mechanical
angul ar velocities ar e different.)
An equation for th e v th ha rmoni c of the rotating magneti c
flux density wav e pro duced by t he field winding , referred
t o a station ar y reference axis, may be derived from
Eq. (26-3) for the same ha rmo nic ..\ ~~llIne that at t = 0 the
axis of th e winding rotating at mechan ica l angular velocity Q run s along the refere nce axis (see Fig . 2G-5a). On this
21 - 0169


Part Two. Energy Conversion by Electrical Machine,

assumption, the angular coor dina tes of an arbitrary poin:

relative to the winding axis, Yo , and relative to t he re ferene,
axis, 1', at an arbitrary instant of time will be connsctso
by an equation of the form

'\' = ~o

+ Qt

Consid ering the above equation together with Eq. (26-3),

Winding axis

W inding axis


St ati qnary


ax es



tri p


(a )

Fig. 26-5 Hi gher harmonics of th e excitation field (a) at t

an arbi trary ti me t

0, (b) at

we obtain the equation for the vth harmoni c of t he ro t ating

flux density wave

B; (a, t) = B v m cos (vpQt - vp,\,)

= B v m cos

(CDvt - va)


Outwardly , Eq. (26-5) is the same as Eq , (25-10) or (25-20)

for t he vth h arm oni c of the flux density wave produced by
a polyphase winding . The coeffi cien t of I' in this equation
is the number of pole pairs for the harmonic in que stion,
vp= Pv' The coeffi cien t of t is t he el ect r ical angular velocity of the harmonic, vpQ = Pv Q = CD v ' The ratio of the
two coefficients is th e me chanical angul ar velocity
vp Q /vp = Q

tho 27 Flux Linkages and EMF's


Flux Linkages of and EMFs

Induced by Rotating Fields


Introductory Motes

When energized, the windings of an electrical machine set

up magnetic fields varying in time and space . As has been
shown in Chapters 25 and 26, the air gap magnetic fl ux
density, no matter how it is produced, can be expanded into
a Fourier series and presented as the sum of rotating fields
differing in the peak va lue of the r adial component, E vm,
the number of pole pairs, Pv' and the mechanical angular
velocity, Q v '
An import ant problem in the theory of electrical machines is to determine the flux linkages with, and the emfs induced in, the phase winding by the rotating fie lds . Because
polyphase wind ings and rotating fi el d windings are always
designed so that the h igher harmonics rapid ly dim in ish in
amplitude with increasing order, the winding fie ld can, to
a good approximation, be represented by the first term
(v = 1) of the Fourier series. The flux density wave of such
a rotating field , with a peak va lue E 1 m , is shown, for example, in Figs. 25-9, 25-13, an d 26-1 .
Re lative to a stationary reference ax is, the flux density
of the forward rotating field is given by Eq . (25-18) as

E (ex , t ) = Elm cos (wt - ex) = Elm cos (pQt - py) (27-1)
Th e emf in duced in a phase winding by a rotating field
can be found as the sum of the emfs in its coils. Therefore,
we shall begin by finding the flux linkage and emf for
one coil.

The Flux linkage and ~ MF of a Coil

Consider a coil displaced from the origin 0 by a distance Xc

along the per iphery of the cor e. The axis of this coil is turned by a me chanical ang le Yc = xci R from the stationary
reference axi s . H ere, R = xpls: is the mean radius of the
air ga p periph ery (Fig. 27-'1). In the general case, the coil
pitch Yc is taken to be shorter than the pole pitch 't o The me2 1*


Part Two. Energy Conversion by Elec trical Machines

chanical angle spanned by the coil or the coil pitch angle

"(y = yc/ R
The rotating wave of flux density described by Eq . (27-1)
travels relative to the coil at mechanical angular velocity Q.
At time t, the axis of the rotating field is displaced from
the reference axis by an angle Qt and takes up the position
shown in the figure. The radial component of B at any point
Fi eld axe s

Coil axi s

Statiqn ary

Fig. 27-1 To determining the flu x linkage of a coil t urn

on the circle, with all angular coordinate "( relative to the

reference axis and at t ime t , can be found by Eq (27-1).
The magnetic flu x <l> links wi th the coil t urn s through an
area A y of a cylindrical surface of radius R; it sp ans an
arc Yc and extends along the generator of the cylinder for
a distance equal to the axial gap length, lo (see Sec. 23-5),
or mathematicall y,
cD =


B n dA = .\ dcD

Rec alling tha t over t he axial gap length t he flux density

at the ax is of the machine rem ains constan t and t ha t in
a cylindrical syst em of coordin ates t he normal component
at a cylindrical sur face is equal to t he radial component,

Bn = B R = B


Ch. 27 Flux Linkages and EMFs

we may replace integration over a surface by integration

over a circle on which the position of a point is defined by
the angular coordinate y.
An elementary area dA may be expressed in terms of an
elementary length along the circle, dx = Rdy, as follows:

= lodx=loRdy

Then, an elementary flux will be given by

= BloRdy


and the integral will have to be taken over the coil pitch,
that is, from y~ = Yc - yyl2 to y~ = Yc "Y,,12:


el>(t=const) =

J de!) = JBhR dy






B 1m i.u
SIll (py -

rot) I'l'~


Upon substituting the limits of integration and expanding

the sines of the sum and difference of angles, namely

rot) +a y/21

sin [(ac -

rot) - a y/21

sin [(a c

we get

el> = el>ym cos (rot - a c ) = lcpel>m cos (rot - a c ) (27-2)


lcpcP m = maximum flux that can link with

a given coil of coil pitch Yc

= ~ -rl"B 1 m = maximum flux linking with a fullpitched coil, Yc = -r


sin (a!l/2) = sin (ycn/2T:) = pitch factor for the

. of the. field


Part Two. Energy Conversion by Electrical Machines

a!J = PY!J = ycnh; = electrical angle spanned by

the coil
a c = PYc = ,Tcnh; = electrical angle defining the
position of the coil axis relative to the reference
axis (origin)
It is seen from Eq . (27-2) that the flux linking the coil
turns varies with an angular frequ ency w = pQ, equal to

Fig. 27-2 The effect of pitch-shortening (chording) on the maximum

flux linking the coil

the electrical angular velocity of the wave . The frequency of

the flux is given by
I = w/2n = QpI2n
Accordingly, the time period of the flux is
T = ill = 2nlQp = yp/Q
It is also seen from Eq . (27-2) (see Fig . 27-2 as well) that
the flux linking a turn passes through a positive maximum
cD = <P!lm when wt - a c = 0, that is, at time
t = ac/w = pyclpQ = yc/Q

when the axis of the field aligns itself wi th the coil axis

t = Q (Yc/Q) = Yc

The amount by which the flux lags behind depends OIl th e

electrical angle a c = PYc defining the position of the coil
relal.ive to the reference axis.
The maximum coil flux is equal to the shaded area in

F. l~.
' -~a :

GIL 27 Flux Linkages and EMFs


The flux linkage of the rotating field with the coil is found
by multiplying the flux defined in Eq. (27-2) by the number
of coil turns We


cD = Wem cos (wt - (X)



is the peak or maximum flux linkage with tho coil.
The instantaneous emf induced in the coil is given by

e = - dWIdt = w1p'em sin (wt - (Xc)

1/2 E e sin (wt- (Xc)

. (27-4.)

The rms value of the coil emf is

Ee =


:2 =


ekpcDmw/V 2


Both the flux linkage and tho emf can be portrayed on

a time vector (phasor) diagram (Fig. 27-3) as complex functions We m and Ee whose pro(+)
jections on the real axis of the
complex plane aligned with
the time axis give the respective instantaneous values:

(' c)
Re {Wem exp [j (wt - (Xc)])

'If = Re

e=Re(V2Ee )



The positions that the above
phasors take up in Fig. 27-3
correspond to the magnetic field shown in Fig. 27-1. Here,
W > 0, because the flux is directed with the coil axis, whereas e < which implies that it is directed against the positive
direction in the coil, in accord with the right-hand screw
The coil emf is

exp (j (wt -

(Xc -

Fig. 27-3 Phasor diagram 0

the coil flux linkage and emf



Pa rt Two. Energy Conversion by Electrical Machines


The Flux Linkage and EMF

of a Coil Group

Each pole pi tch of a double-layer winding has q coils of

a given ph ase (in Fig . 27-4, q = 3).
Th e wav eform of the flu x linkages and emfs for the coil
group shown in Fig. 27-4, plotted by Eq. (27-3), (27-4) or
(27-6), appears in Fig. 27-5. Because the coils in the group
ar e displaced from each ot her by an electri cal angle
CG z

Pl' z

(tzh:) rt = CG C 2


CG C 3

= . ..

CG C 2

the flux linkage and emf ph asors are likewise displaced

from each other by t he same angle .
1)th (say, second) coil lag behind
The events in the (k
those in the kth (say, first) coil by the t ime required for the
flux density wave to mo ve through a mechanical angl e
I'z, t ha t is ,




This lag must be all owed for in combining the flux linkages (and emfs) within a gi ven coil grou p.
The coil -group flu x linkage and emf phasors, 'g and Eg ,
are each the ph asor sum of the coil linkages and emfs,
\fc l ' lfC2 ' 1fc 3 and ECl> Ec2' EC3. Going back to
Eq. (24-29) and Figs. 24-10 and 24-11 in Sec. 24-5, it
will be recall ed that the problem of combining several
phasors equal in magnitude,
'c lm


= 1c2m =

1 c3m

'f c m

= E C2 = E C3 = E c

and displaced from each other by the same angle CG z has

already been sol ved in determining the mmf of t he win ding.
Therefore, the coil-group flu x linkage and emf may be
wri tten
'I' g m = q1cmk ct = qWck pkctfP m
E g = qEck ct
The coil-group flu x linkage phasor is directed alon g
t he ax is of symmetry of the coil ph asors and, as is seen
from a comparison of Figs. 27-5 and 27-4, is tur ned
through an angle (rot - CG p h) from the real axis of the
complex plane. H ere, CGph = Pyph a is the e~ectri9Rl
~ ,


Ch. 27 Flux Linkages and EMFs

Fig . 274 EMF induced in a coil group


Fig . 27-5 Phas or dia gram of flux linkage and emf for a coil grou{J


Part Two. Ene rg y Conversi on by El ectri cal Machines

angle of a coil group or the ph ase axis. It is to be noted

that t he ax is of a coil group (the ph ase ax is) is the axis of
symmetry for the coil group . The an gle defining the position
of th is axis is foun d as the ar ithmetic mean of t he angles
defini ng th e positions of the coils in the group
"'Ph =

YCl +'\'C2+ + YCq

o.C1 +o.C2+ +'Xcq


aph/ P

The coil-gro up emf lags behi nd th e coil -gr oup flux linkage
by n /2, and ma y be written as

The Flux Linkage and EMF of a Phase

A phase of a winding is ma de up of coil gr oups connect ed in

series-parallel (see Sec . 22-3) A ph ase of a double-la yer
winding ha s 2p id en t ical coil groups (one group per pole
Group II


if ('f)

Group X

Group II


F ig . 27-6 EMF induced in th e coil


Group X


groups of a ph ase by a rot ating

pitch). As an exa mple, Fig. 27-6 shows t he coil gr oups of

pha se A in a four-pole, three-phase win ding (2p = 4). Its
comp le te circu it diagra m is sh own in Fi g. 22-5 . A d J a ceI~~


Gil. 27 Flux Linkages and EMFs

coil groups in t he ph ase are displaced from one another by

one pole pi tch 1: or by a h alf of the pole-pitch angle

,,? p/2 = 2rrJ2p = nip

The respective el ec tri cal angle is
r:t. p /2

= p,,?p/2 = n

Therefor e, the flux linkages and emfs of the backward coil

groups in the sam e phase, 'ijf gXm and if g X, are in antiphase with t hose of the forward
coil groups, 'ijf g Am and Jj; g Ao
If a phase h as a parallel pa ths
(circuits) , t hen each pa th
cont ains 2pla coil groups. T11e
forw ard coil gr oups are connected aiding (with their finishes t o t he phase finish), wh ereas the backward coil groups are
connec ted in opposition (with
t heir starts to the ph ase fi nish).
Exactly this form of connection of coil groups in parallel
paths is sh own in F igs . 22-5
an d 22-6 . Now the positive direction around a parallel path
Fig. 27-7 Ph asor di agram of
(from its fi ni sh X to wards its
flux li nk age and emf for a phase
start A ) is t h e same as the
in a doub le-l ayer winding
positive direction around a
coil gr ou p (fr om its fin ish F t owards its star t S) in all t he
forward gr ou ps connected a iding (A) and is opposite to the
positive direction around all the backward coil groups (X) .
With this arrangeme n t, t he flux linkages and emfs of the
coil gr oups are combined ar ithme t ica lly wi thin a parti cular
path , and the flux linkages an d emfs in all t he paths are the
same (Fig. 27-7) .
The ph ase fl ux li nk age an d ph ase emf are resp ectively
equ al to th e flu x linkage and emf of a path

{if. ph m -_

p' l' g -,1 III -p 'l' g .K m _ ?




gA m

~2 7- t 2~


Part Two. Energy Conversion by Electrical Machines

. The phase emf can be expressed in terms of the phase flux

li nkage directly

Ep h = 2pEgAIa =

- j2PU/ gAmla V 2"


jro'f p h m lV 2


The phase flux linkage (see Figs. 27-6 and 27-7) is in the
same direction as the flux linkage of the main coil group
whose axis is taken as t he phase axis and makes an angle


Eg z




Fig . 27-8 Phasor diagrams of flux linkages and emfs for the phases
and coil groups of a three-phase winding:
(a) for two paths (circuits) in a phase; (b) for four paths (circuits) in
a phase

= PYph with the ori gin. The phase emf lags behind
the phase flux linkage by rrJ2 (see Figs. 27-7 and 27-8).
The magnitude of the phase flux linkage or ph ase emf is
2pla times the magnitude of the coil-group flux linkage or
emf. The ph asor diagrams in Figs. 27-7 and 27-8 are plotted
for phase A consisting of the coil groups shown in Fig. 27-6.
In Fig. 27-6, the number of paths is a = 1, so 2pla = 4 (see
the dashed connections in Figs . 27-6 and 22-5) . In Fig. 27-8a,
the number of paths is a = 2, so 2pla = 2. In Fig. 27-8b1
G = 4, so 2pla = 1 (see Fig. 22-6b) ,


Ch. 27 Flux Linkages and EMF's

In accord with Eqs. (27-8) and (27-11), the peak value of

the phase flux linkage is

1phm = 2p1gm/a = wkwCP m

w = 2pwcq/a
k w = kp/k d


= number of series turns per phase

= phase winding factor (for the funda-

mental component of the field)

cD m = peak value of the magnetic flux over
a pole pitch
The rms value of phase emf given by Eq. (27-13) is
E ph = ro1PhmlV2= 2njwkw Cl)m/V '2


The Flux Linkages and EMFs

of a Polyphase Winding. A Space-Time
Diagram of Flux Linkages and EMFs

All the phases in a symmetrical polyphase winding are

identical in arrangement. Adjacent phases, say, phases A
and B, whose axes make mechanical angles I' A and I' B wi th
the stationary reference axis, are displaced from each other
by a mechanical angle (see Fig. 22-6)
I' BA = I' B - 1 ' A = Zsilmp = I'p/ m
or by an electrical angle



= PI' BA = 2n/m

Therefore, the phase flux linkages and phase emfs are the
same in magnitude (Fig. 27-8):

1 Am = 1 Bm = 1 Cm = 1m

In the case of a forward rotating field, that is, one moving

from phase A to phase B to phase C, the flux linkages and
emfs of a polyphase winding form on the complex plane an
m-ray star in which the adjacent arms are displaced from
each other by an angle 2n 1m (for a three-phase winding, this
angle is 2n/3, see Fig. 27-8).
Let the axis of phase A run along the stationary reference
axis. Mathematically, this will be written as

=PI'A = 0


Part Two. Energy Conversion by Electrical Machines

the instantaneous phase flux linkages will be

1p" A = 1p" Am COS (wt - C(.A) = 'f A m COS tat
1]1 B= 1p" Bm. COS (wt - CXBA) = 1p" Bm COS (wt - 2:n:/3)
lJF C = 1p" c m COS (wt - CXCA) = 1p" Cm COS (wt - 4:n:/3)

Or, in complex notation,

= Re

1p" A
1p" B =
1p" C



[W Am] =

["If Bm] = Re

[lY em]

= Re


["0/A m

exp (jwt)l

[1.4.m exp (-j2:n: /3)]

[1f Am exp


The instantaneous phase emfs can be written in a similar


v:2 E cos (wt - :n:/2)

= V:2 En cos (wt- :n:/2-cx nA )
= V:2 E cos (wt - :n:/2 - 2:n:/3)

eA =

Or, in complex notation,



en = Re

lV2 EA ] =

rv :2E


Re {V2 EA exp [j (wt - n/2)]}

= Re

[V 2" En exp (- j2n/3)]

By analogy with the phase currents (see Sec . 25-3 and

Fig . 25-6), the phase flux linkages (phase emfs) can he depi cted on the complex plane of a two-pole model as complex
functions common to all the ph ases .
For the three-phase winding whose flux linkages and emfs
are shown on the time complex plane (Fig . 27-8a), the
flux linkage phasors

-qr m =

1p" m

exp (jwt)

and the emf phasors


E p h exp [j(wt - :n:/2)]

corresponding to the respective phase quantities are shown

on the space complex plane of the model in Fig. 27-9. In
the two-pole model, the phase windings are each shown for
clarity as a single coil; the positive direction is shown in the
sectional view drawn in the same figure. The phase axes are
drawn through the centres of the coil groups represented by


Ch. 27 Flux Linkag es and EMF's

one coil. T he instantaneous ph ase flux linkages (or the

ins tan taneous ph ase emfs r educed by a factor of 11 2) are given by pro jections of the respective phasors on the axi s
of t he respective phase.
Because the position of the flux linkage 01' of the emf
relative t o the ax is of a given phase in Fig. 27-9 is the same
as tha t of th e flux linkage (01' emf) of that ph ase relative to
the real axis of t he time complex plane, their instantaneous
flux linkage (01' the ins tantaneous emf) is the same in either case.
The sp ace complex plane in
Fig. 27-9 also shows the complex funct ion

13 1m

= E lm

exp (jwt)

depicting the magnetic flux

den sity of t he rotat ing fi eld
we are considering [see Eq .
Fi g. 27-9 Rot ating-field flux
density, ph ase flux linkages and
(27-1)]. It has been plotted in
phase emfs shown on th e space
exactly t he same way as in
complex pl ane of a two-pole
Fig . 25-10. (It will be recalled
tha t the axis of ph ase A has
been assumed to run alon g the stationa r y reference axis .)
As follows from Fig. 27-9 and the applicable equations,
the complex functions depi cting the m agn etic flu x densi ty
of a rotating field and the flu x linkage pro duced by th at
field are both in t he same direct ion . This is because the
phase flux linkage is a maximum at t he instant when t he
magnetic flux density at t he phase axis is a maximum
(see abo ve).
Axis c


the Flux Linkages and EMFs due to

. the Harmonics of a Nonsinusoidal
Rotating Magnetic Field

As h as been explained in Chapters 25 and 26, a rotating fie ld

ma y, in addition to t he fun dam ental com pone nt, con tain
an amount of harmonics .
A rotating field cont aining harmonics is nonsinusoidal. The flu x linkages and emfs produced by the


Pa rt 'Two. Energy Conversion by Electrical Machines

harmonics can be fou nd by the equations deri ved for the

fundamental component, if they are re-arranged to include
the respec tive harmonic qu antities, such as B vm, 'Lv, and Q v '
From Eq. (27-14.) it follows that the vth harmonic component of a rotating field gives rise t o a flux linkage wi th each
phase winding, defined (the peak value is meant) by
"Ifphvm =

where k..vv =



:;= phase
winding fac tor for the vth
k p v = pitch factor for the vth harmonic, Eq. (24.-27)
k d v = distribution factor for the vth harmonic,
Eqs. (24.-32) and (24-33)
cD vm = (2/1)'L v l oBvm= magnetic flux due t o the vth
harmonic over a pole pitch, Eq. (27-2)
It follows from Eq. (27-15) that the rms value of the ph ase
emf induced by the vth harmonic of the magnetic field is
given by

kpvk d v

E ph v = (wv/ V 2) cD ph vm = (21lV

= 2 VZ!vwkwv('L/'V)


loB vm

2) !vwkwv(Pvm

= QvPv = Qvpv = electrical angular velocity of

the harmonic, equal t o the circular frequency

of the induced emf
! v = w)21 = fr equency of t he induced emf
The emfs induced by the harmonics are superimposed on the emf induced by the fu ndamental component and
. affect the resultant phase emf and, in the final analysis, the
performance of the m achine. This effect va ries accor ding as
the nonsinusoidal rotating field is produced. Consider two
cases which are most typical of all , namely: t he magnetic
field produced by a polyphase winding (see Chap. 25), and
the magnetic field produced by a rotating fie ld wind ing
(see Chap. 26).
1. Typically, t he waveform of the nonsinusoidal ma gnetic field set up by a polyphase win ding is continually varying in sh ap e, because its rotating h armonics tr avel at different mechanical angular velo cities (see Fig. 25-13 an d
Sec. 25-6):
Q.. .

cslp ;

lp v

where ro = 211 is the circular fr equency of t he currents

in the polyphase win di ng.

ci, 27 Flux Linkages and EMFs


It is readily seen [see Eq. (27-17)] that the rotating harmonics of this field induce emfs of the same frequency
equal to the frequency of the current in the winding

CO v





A more detailed analysis would show th at . the emfs due

to the harmonics are in phase with the fundamental emf
and are added to it arithmetically. The harmonics do
not affect the waveform of the fundamental emf, and the
resultant emf is sinusoidal.
The effect of the vth harmonic on the rms emf depends on
the ratio



Therefore, even for a concentrated (q = 1), full-pitched

(Yc = 't) Winding, when kwv = k w = 1, this effect is 1/v of
that produced by the vth harmonic of magnetic flux density
(with a peak value B vm) on the fundamental flux density
(with a peak value B l m).
Still, the total emf induced by all the harmonics



E phv=2 V2f w'tl.-!




may be fairly large in magnitude, especially in a fullpitched winding and with small values of q.
In practical machines, the ratio
Ea,rmslEph =



EphvlEph =




may range anywhere between 0.005 and 0.05. The smaller

values apply to short-pitched windings for which Yc ~
~0.83't, and q~1, so that k wv/vk w <i:..1 (see Chap. 24).
Energy conversion by electrical machines mainly utilizes
the fundamental component, the sole contributor to the useful
field. The fields set up by the harmonics may be classed as
leakage field. In Soviet usage, they are referred to as differential or difference leakage fields, because their sum may be
looked upon as the difference between the resultant field of
a polyphase winding and the fundamental (or useful) field.
The total harmonic emf
','" E phv
E a,rms = L..J

2 2-0169



'I'wo, Energy Conversion by Elect rical Machines

is accordingly called the differential (or difference) leakage emf

and treated separately from the emf induced by the fundamental field (see Sec. 28-7).
2. For the nonsinusoidal magnetic fi eld esta blished by
a rotating fi eld winding, it is characteristic that the magn etic flux density waveform remains unch ange d as t he fiel d
rotates (see Fig. 26-5 in Sec. 26-3).
Because of this, all of its harm onic s rotate at the sam e
mechanical angular velocity equal to the mechanical angular
velocity of the field winding

The frequency of the emfs in duced by the fi eld harmonics

is propor tional to the order (number) of the harmonics
[see Eq. (27-17)]






where co = 2rr,f is the circular frequency of t he emf induced

by the fundamen tal fi eld.
Thus, the emf induced by t he vth sp ace harmonic of the
field is the vth time harmonic of the emf. The contributio n
by the var ious harmonics depends on t he ra tio
Eph ,,/E ph

= kw"B vm/kwBlm

(27-18) -

where, as will be recalled, t; = v] , The higher values of t he

above ratio correspond t o a lar ger departure of the result an t
emf from the sinusoid al waveform . On the other hand, for
energy conversion b y elec trical machines an d transform ers
t o be most economical (to suffer a minimum of loss) , it is
essential that the voltages , emfs and currents involved be as
close to a sinusoidal waveform as practi cable. One of the
causes of the increased losses associated with a nonsinusoidal voltage waveform is the circulating currents pro duced by
harmonic emfs and flowing between t he machines when several of them are connected for parallel operation.
In designing an electrical machine, ever y effort is ma de
to make the winding volt ages as close t o sinusoidal as
practicable . In Sovie t pr actice, this is assessed in terms
of the devi ation factor of a voltage (current) wave defined as

l/ >;
k (per cent ) =

v=l= l




Cit. 27 Flux Linkages and EMF's

where Elm is the peak value of voltage at the fundamental

frequency, and E vm is the peak value of the vth harmonic
One way to achieve this goal is to make as sinusoidal as
possible the waveform of the magnetic flux density due to
the excitation field (we have already shown how this can be
don e with a concentrated and a distributed fiel d winding in
Chap. 26). Still, for all the measures taken, the distortion
factor of the exci ta tion field may exceed the limit. In fact,
if the tim e wavefo rm of the emf were allowed to follow that
of magnetic flux density in space, the machine would not be
able to perform its designated function. Fortunately, this
only happens (compare Figs. 27-10a and b) in a concentrated
phase winding with one coil pel' group (q = 1) and wound
wi th a full pitch (Yc = 1:). In the circumstances, kw = kwv
= 1 and, as follows from Eq. (27-18), the ratio of the
harmonic emfs to t he fundamental is the same as the ratio of
t he harmonic flux density to t he fundamental component,
or mathematically
Eph v/Eph = Bvm/B l m


When t he phases of a three-phase winding are st ar-connected, the line volta ge is free from triplen harmonics of emf,
t hat is , t hose for which v = 3k = 3, 9, 15, etc. , where
k = 1, 3, 5, et c. This is so becaus e (see Part One of this
te xt) t he ha rm onic emfs of such an order are the same in all
the ph ases (eA V = env = ec v), and cancel one another in
the line emfs found as the difference of the phase emfs:

eAR, ~ eM - en> ~ 0
When a t hree-phase winding is delta-con nect ed, the line
voltage is again free from the triplen harmonics but for
a differen t reason (see Par t I of the t ext). The poin t is that
around a delta circuit t he t riplen harmonics are added
together arithmetically, giving rise to a circula ting current

I v = 3E Avl3Z p h

so that the respecti ve harmonics of line voltage add to zero:



= 0

Thus, as we have seen, the waveform of liue voltage in

a three-phase winding is impro ved as (:0 111 pared with the


Pari Two. i!;iiergy Conversion by Electrical Machin es

waveform of magnetic flux density (see Fig. 27-10c) even in

the case of the least perfect winding configura t ion (Yc = 't ,
q= 1). A further improvement in the wave form of bo th phase
and line voltage in t hree -phase windings can be obtained



Fig. 27-10 Effect of three-ph ase winding arra ngement on the waveform of phase and line emfs:
(a) waveform of the excitat ion-field flux densi ty; (b) phase emf for
Yc = .., q = 1; (c) line emf for Yc = .. and q = 1; (d) phase emf for
Yc = 0.83.., q = 2; (e) line emf for Yc = 0.83r , q = 2

by using dis tribut ed windings ts > 1) woun d with a shor t

pi t ch (Yc ~ 0.83"1"). In such windings, for all harm onics, except the too th harmonics (slot ripple), as has been
shown in Figs. 24-9 and 24-12, we get

kw vlkw = (k pvkdvlkpkd) ~ 1

Ch. 28 Ind uct anc es of Polyphase Windings


Ephv/Eph =

kwvBvm/kwB lm ~ Bvm /B 1m

This implies that t he waveform of emf is more sinusoid al

than t hat of magnetic flu x den sity (see Fig. 27-'1Od an d e).
As is seen , when Yc ~ 831: and q?;3:- 2 , the emf is practically
sinusoidal , even though th e waveform of magnetic flux
density due to t he exci t ation field is substantia ll y nonsinusoidal.
It should be add ed t hat in such windings the rms value
of phase or line emf does not practically diff er from the rms
value of t he fun da mental emf
E ph Z= -.V,rE 2 + E 2ph 3 + E 2p h 5 + E 2ph7 + . . ,....,
,...., ph


T he Inductances
of Polyphase W indings


The Useful Field

and the Leakage Field

Let us consider the magnetic field in an elect ric al machine

with two pol yphase windings one of which is wound on the
stator, and the other on the roto r .
Assuming that the relative permeability of the stator and
ro tor cores is infinit ely large (u, = 00) , the st ead y-sta te
magnetic field of such a ma chin e can be visuali zed as consisting of two components , namely t he us eful field an d the
leakage field.
As will be recalled the useful m agnetic f ield is t hat which
is associated wi th t he fu ndamental component of t he ra dial
magnetic flux density in the air gap . This field plays the
decisive part in energy conversion . When fLr = 00, the useful field may be im agined as form ed by tw o fi elds which are
stationary rel ative t o each other, namely the useful stator
field set up by the currents in the stator winding, and the
useful rotor field set up by the currents in the rotor winding.
Of course , the ai r gap flux density due to each of t hese field s
contains onl y t he fund ament al componen t . In turn, the useful
stator (rotor) field may be visualized as the sum of the usefuJ
fields established by the various phases of the stator (rotor)


Pa rt Two . Energy Conversion by Electrical Machines

The leakage f ield is that which is es tabli shed by the sets

of currents in t he s tator and rot or windings tha t do not contribute to th e useful field . In other word s, when the fu nd amen tal fluxes of the st ator and ro tor fiel ds cancel out , the
leakage field only exists in the machine .
The total flux linkage of a polyph ase winding may he
visualized as the sum of the useful flu x linkage and th e
leakage flux linkage. Th e form er is associated with the useful
fie ld whose lines close via the air gap and link both windings of the machine . Forithis reason, it is called the mutual
field . The leakage flu x linkage is associa te d with that par t
of the leakage field which links only one (stator or rotor)

The Main Self-Inductance of a Phase

The main phase self-inductance is associat ed with the mutual

flux linkage produced by the respective phase current. Let
us find the main self-inductance of phase A in t 11f' sta tor
winding. Suppose that the phase winding carries by ~ nositive current whose peak value is

=V 2/ A

In Fig. 28-1, phase A is shown for clarity as a single coil.

The fundamental component of t he phase mmf with the
peak value given by Eq. (24-28) is

F phlm = 2 V 2 / AWlk Wl!np

In the air gap, it gives rise t o a cosi nusoidally distributed
mu tual magnetic field whos e flu x densi ty at the phase ax is ,
according to Eq. (25-18), is

1m -


ph l m

'A 0-

V2 rrp6k
I Aw1kw1llo

The fund amental component of the phase mmf and t he

norm al component of the air gap flux density are sho wn in
Fig. 28-1a, and the ph ase magnetic field pa ttern in Fig. 28-1b.
Rec alling that t he axis of the magnetic fi eld runs along
the phase axis , its flu x linkage with the phase winding [see
Eqs. (27-2), (27-13), and (27-14)] can be written as
'I'AAm =

wl k wl<1>m = - T:l!'J WlkwIB lm



Gh. 28 Inductances of Polyphase Windings

(.4"R This flux linkage is proportional to the number of mutual

field lines that cut the surface which spans the contour of
coil AX representing phase A actually consisting of
many coils.
Axis If

Axis A

(a )


Fig. 28-1 The mutual magnetic field of phase A in a two-pole model:

(a) distribution of the normal component of the phase flux density;
(b) magnetic field pattern for phase A

By definition , the main self-inductance of a phase A is

)2 -r l 6
L AA = ur
r AAm [i~A = -4~lO
- 2 ( Wi k Wi"""--k

u 0

- '1)

It is seen from Eq. (28-1) that the main self-inductance of

a phase depends on the air gap dimensions (lo, 't, 6, k 0),
the magnetic properties of the air gap (/lo), and the characteristics of the stator winding (p, w, kW 1 ) ' In our example, the
air gap is uniform, so the main self-inductance is independent of the relative position of the rotor and stator.

The Main Mutual Inductance between

the Phases

The main mutual inductance between the phases varies with

the electrical angle between the phase axes. To find the
main mutual inductance between phases A and B of a poly- .
phase winding, with their axes displaced from each other by

3<1 4

Part Two. En e rgy Conversion by Electrica l Machines

an electri cal ang le a BA. = PYB A, we should first find the

flux linkage between t he useful field of phase A (shown in
Fig. 28-1a and b) and th at of phase B. It has been shown in
Sec. 27-2 that the flu x linkage of a rotating field with a phase
is proportional t o the cosine of the electrical angle between
the field axis and the pha se axis. (In Eq. (27-13) and in
Fig. 27-7, this angle is ro t - ap h.) Therefore , the flux linkage of the phase A fi eld (for i A . = V 2 I A) with that of
phase B is

'P'BA m = 'P'A A m


By definition, the main mutual inductance between phases

A and B is
L E A = lJf B Amli A = ('P'A A.m1i A) cosaBA
= L A. A. cos a BA
For the three-phase winding in Fig. 28-1,
a B.'!.

Therefore ,


a CA


cos aB A = cos a C A = -1/2

and the main mu tual inductances between the phases are
L B A = L C A = - L A A I2
It is seen from Fig. 28-1 b that the plane of the phase B
coil is cut by half as many field lines as t he plane of the
phase A coil. Also, whereas the plane of the phase A coil
is cut by fi eld lin es in t he positive direction (with the axis
of phase A ), the plane of the phase B coil is cut by field
li nes in the nega tive direction (against t he axis of ph ase B).
This difference in flu x linkages controls the magnitude and
sign of t he mu tual inductance.

The M ain Mutu al Inductance Between

a stator Phase and a Rotor Phase

As in the previous section , this mu tual inductance is a fu nct ion of the cosine of the elec trical angle between the axes of
the stator and rot or ph ases considered. Also, in finding t he
flux linkage of primary (say, stator) phase A (the primary
phases have upper-case letters in their indexes) with, say ,
secondary (say, rotor) phase b (the secondary phases have
lower-case let t ers in their ind exes) , it . is important to re-

Ch. 28 Inductances of Polyphase Windings


member that a secondary phase h as a different numb er of

turns, w 2 , and a different winding factor , 1~w2 '
If the electrica l angle between the axes of phases A and b
(see Fig. 28-1) at a given instant is equal to CX b A' then the
flux li nkage with phase b is given by

Accordingly, the ma in mutual inductance between ph ases A

and b is
wher e
k ) 'tl(\

= -P1C".-

( W 1/~W 1 w'" "-k

U Ii

is the peak value of mutual inductance between a pr imary

phase and a secondary phase (say, phases A and a, when their
axes coincide and t he electrica l angle between them is zero).
As is seen from Eq. (28-3) , when the rotor rotates at
a 'mechanical angular veloc ity Q , the angle CXbA = Qpt
inc reases in a linear fashion, and L b A is varying harmonically . In Fig. 28-1b, the mutua l inductance is positive, because
the field lines cut th e coil pl an e by in the positive direction
(along its axis).

The Main Self-Inductance

of the Complete Winding

In add ition to the self and mutual inductances examined

in the previous sections, which are found by definition, it is
convenient in the theory of electrical ma chines to introduce
the concept of the self -inductance of t he complete winding.
It can be defined as the self-inductance of a ph ase (say ,
phase A ) which is in tur n defined as the ratio between the
ma ximum flux linkage due to all t he primary phases with
phase A , and the peak value of the phase A cur rent * . For
a three-phase winding,


=WAmIV21A =

'Y A Am+ABm+WAcm
, I_ ~


* This self-inductance is th e same for balanced sets of PPS and

NPS currents, but is different for ZPS currents.


Part Two. Energy Conversion by Electrical Machines

On expressing the flux linkages in terms of currents, main

self- and mutual inductances,

1V AA m = V'2"JA L A A , 1jf ABm = V 2I B L A B , 1VA C m = V 2Jc L A C

and recalling that, in accord with Eq . (28-2) ,


= -



and also no ting t hat for balanced sets of PPS and NPS currents
IB+l c = - I A
we can see that the m ai n self-induct ance of t he complete
winding can be written in terms of t he main self-inductance
of a phase as

L I1 = ZL A A

= (6Ilo/p:n;2) (w 1kw 1)2 (Tlo/ako)

In the .general case, for an Tnt-phase winding, the main

self-inductance is

L I1 -:- (Tnt/2) L A A

(2TnlIl0/p:n;2) (WtkWl)2 (Tl r,/8k{j) (28-4)

As an alternative , the main self-induct an ce of t he com plete

winding can be fou nd from t he peak flux linkage of the
fundamental ro ta ting field set up by all the ph ases , wi th one
of the phases. In accord with Eq. (25-9), the peak value of
the fundament al mmf of the Tnt-phase prima ry win ding that
sets up the field is

_ mt If'Z IAwt kwt


1(O m -

and, in accord with E q. (24-17), the peak value of magn etic

flux density is
B t (l ) m

~to Flm/6k{j

As follows from the above equations, t he peak flux linkage

of a ro ta tin g field wi th a phase is

'Y..l m = (2/:n;) Tl {j lV 1kw 1B

1(l )m

and the main self-inductance of the primary winding is

L I1 = ' Am/V 2 t ,
Naturally, the result -is t he same as that given b y E q. (28-4):

Ch.:28 In ductanc es of Polyphase Windings



The M ain Mutual Inductance between

a Primary Phase and the Secondary

Win ding

The main mutual induct ance b etween phases of different

windings is likewise found from the peak flux linkage of
all the second ary phases (or , in other words, due t o t h e
ro tating seconda r y field) with a primary phase. It is equal
t o the ratio of t h is flux linkage to the peak value of secondary current. The flux linkage with phase A is a m aximum
when the axis of this ph as e is aligned with that of phase a,
t h e axis of the ro t ating field is aligned with the axis of
phase A , and t he current in ph ase a is at its peak value.
The peak fund ame n tal m mf of the m 2-phase ' secondary
win ding which set s up t he fie l d is giv en by Eq. (25-9)

F l( 2 ) m

(m zl f2/:n: ) (I a W2 kw2/ p)

The peak value of t he associated fundamental flux density is


~LoFI (2)m/5k ~

T he pe ak flux linkage of t he ro t atin g field with the primary

phase A is
'1'Am ~ (2/:n:)T:l oWikwIBI( 2)m
and the main mutual inductance between a primary phase
and a secondary phase is

L i2m = \f' Am/ )/ 2" I a = (2m2~O/p:n:2) (wikwiW 2kw2) (T:l~ /Oko)

I t is an easy matter t o see t h at t h is parame ter is connected
t o the peak mutual inductance between a primary phase and
a secondary phase by a simple relation of the form

L I2m = m 2Lm/2
which . is similar t o Eq. (28-4).
The ma in mutual induct an ce between a se condary phase
and the primary winding is gi ve n by
L 21m

= m]L m/2

and, if mI :;6 m 2 , it differs from the mutual inductance between

a primary phase and the secondary winding, L I2 m .

Part Two. Energy Conversion by Electrical Machines



The Leakage Inductance

of the Complete Winding

By definition (see Sec. 28-2), a leakage field exists when

the fundamental components of the air gap magnetic fields
due to the currents in the stator and rotor windings cancel

Bj (z)m

If, to simplify the argument, we assume that the rotor

winding is stationary relative to the stator winding, that the
axis of phase A is aligned with that of phase a (see Fig. 28-2),
Axis A

IAxis a

BI (I ) m

~1------- ---........

'PUn, , 2m

Fig. 28-2 Leakage magnetic field due to the polyphase windings on

the stator and rotor ('If g stands for 'If t. and 'If line for 'If e)

and that the air gap is uniform , then the fundamental component of the air gap magnetic field will vanish when the
fundamental mmfs of the two windings are equal and opposite in peak value. Mathematicall y, this condition may he
written as

This condition will he satisfied if the secondary phase currents are appropriately related to the primary phase currents


eh. 28 inductances of Polyph ase Windings

The oth er primar y (or seconda ry ) pha se cur rents and th e

ph ase A (or a) current form between t hem a bal an ced set of
PPS or NPS currents.
The flux linkage of t he le ak age fiel d wi th prim ary phase A
s a m aximum when the current in that pha se is a ma ximum,
i A = V2 lA, whereas the current in secondary phase a must
be such that
ia =

i A (mtwtkwtlm2 w 2kw2) =

- V 2" I a

Precisely such currents in phases A and a, an d appropri ate

currents in the other primary and secondary ph ases set up
the magnetic field shown in Fi g. 28-2.
Given a set of curren ts , the leakage field can be fou nd by
electrical-field equa tions (see Chap. 23). Then one finds the
leakage flux linkage with phase A in the various parts
(ll'sl m, 'l'tlm, 'l' el m, and 'l'd 1m), the tot al leakage flu x linkage
with a phase

'Y al m = 'Y sl m+ 1Yum + 'Yel m + 'Y dt m

and the leak age self-induc t an ce of a ph ase

= 1Ya1m ll/ 2" I A

In this way, the leakage flux linkage of a phase is found with
allowance for t he effect of t he other ph ases on t he stator and
rotor. The leakage flux linkage of the secondary winding is
calculated in a similar way

L a2 =

11' a2m l V 2" I a =


+ ll't2m + ll'ezm +'If d zm

V 2I a

The field lines contributing to the slot leakage flux linkage

('l'slm and 'l's2m) , the too th leakage flux linkage ('Yu m
and 'l't2m) , and the coil-en d leak age flux linkage ('l' el m
and 'l' e2m) are sho wn in Fig. 28-2 (see also Chap. 23).
The differential leakage flux linkage ('l'd l m and 'Y d 2 m ) is
also taken in to consideration.
Th e stator and rotor (primary and secondary ) windings
always differ in t he number of phases, the number of slots
per pole per ph ase , and so on. As a result , the stator and ro tor
windin g factors are differen t even for t he harmonic mmfs of
the same order, and th e harmo ni cs themselv es ro t ate at diffe rent angular velocities. This is t he reason why th e harmon ic
mmfs do not canc el one another , although t he fundam ental
mmfs do.

Part Two. Energy Conversion hy Electrical Machines

The differentia l leak age emf, E od ' induced by t he higher

harmonics has been discussed in Sec . 27-6. In it s terms, the
differential leakage flux linkage for the primary winding
may be written
lJfod m =

V 2 E od/2nj = 1/2

~ E phvl2nj


= (2/n ) w(t l{j .:....:

where B vm = EV(1)mflo /6k {j.

Omitting the details, we sh all only give an equation for
the leakage self-inductance of a primary phase, stemming
from Eq . (28-6)

where A{j1 = AS1

Ae1 Ad1 is the permeance for
the leakage flux linkage (a dimensionless quantity).
The terms of the sum above are the permeances of the
various leak age fie ld s, defined per unit of coil-side design
length . The higher a given permeance, the larger the associated leakage inductance . The magnitude of a permeance depends
on the dimensions governing the respective leakage field.
To facilitate computation, equations giving the various
permeances have been developed on making certain assumptions as regards the leakage field pattern and taking the
permeability of the ferromagnetic parts of a machine to be
infinitely large.
For a three-phase, double-layer winding, the various
permeances can, in a first approximation, be fonnd by the
following equations.
(i) The slot leakage permeance (for the rectangular slot
of Fig. 23-4)

= (helb o

+ hr/3b s) (3B + 1)/4


radial depth taken up by conductors in a slot

= clearance between . conductors and airgap


slot. wid th

= width of opening towards the airgap

B = Yeh = chording (pitch-shortening) factor

(ii) The tooth leakage permeance (see Fig . 23-4)
At1 =[1.1 (6'/b o)-O.35(6'/b eF'-O.26] (3~t1)

Ch. 29 Electromagnetic Torque


where B = k oB = effective radial length air gap

(iii) The coil-end leakage perm eance (see Fig.

A.e i


= 0.3 4 11 ~ro.i + 0.1 (~lTql/l0) ~ 0.3~lLql/l0

where ~ro'l

= lro.l/Ycl =


lro .l/~lL

= rela tive coil-end overhang

= ch ord ing (pitch-s hor tening) factor

(iv) The differential leak age perm eanc e (see Fig. 24-3)
A.d l = (0.7 to 1.0) (tz/12 B k o)
where t z = to oth (slot) pitch
B = radial air gap length
The factor 0.7 to 1.0 in t he equation for A.dl depends on the
degree of pitch shortening (chording), relative slot opening
(bo /t z and bo/B), the damping effect of currents induced in
the secondary winding, etc.
For the secondary three-phase winding, the above permeances can be found by the same equations on replacing
the in dex "1" with "2".


The Electromagnetic Torque


The Torque Expressed in Terms

of Variat ions in the Energy
of the Magnetic Field

Let us consi der an in duction or a synchro nous a .c. machine

wi th a un iform air gap. ' Ve shall repl ace the t oot hed cores by
smo oth ones an d in tro duc e a n equivalent air gap leng th
Bo = ko B
where k 0 is the slot fa ct or acco unting for t he effect of cor e
saliency on the permeabili ty of the air ga p [see Eq. (24-10)J,
Suppose that the stator is wound with a symmetrical polyphase winding with In l
2, an d that t he ro tor carr ies
2, or
ei ther a symmetrical pol yphase winding with 1n 2
a single-phase field wind ing (this applies to a synchronous
machine). Let the stator winding carry a set of PPS currents ,
II, varying with an angular fr equency WI' and the rotor
carry either a set of PPS currents 12 (in the cas e of a polyphase
winding) varying with an angular fr equency W 2 , or a direct




Part 'Two. Energy Conversion by Electrical Machines

current 1 2m (in the case of a single-phase field winding), for

which WQ = O.
As has been shown in Sec. 21-2, such a machine will be
capable of unidirectional energy conversion only if the
frequency of stator currents, WI' of rotor currents, W 2, and of
mutual inductance, W = pQ, satisfy a certain condition.
More specifically, it is required that
WI + w2 = W = pQ

If this condition is met (to make the matter more specific,

let W < WI and W2 = WI - i), then, as can readily be
shown, the fundamental components of mmfs (or of the
rotating fields) due to the balanced sets of currents in the
primary and secondary windings will be rotating relative to
the stator at the same mechanical angular frequency
0. 1 = Wl/P

Referring to the two-pole model of a machine (Fig. 29-1)

whose polyphase (three-phase) windings carry PPS currents,
it can be seen that the fundamental mmf of the primary

or the fundamental component of the air gap magnetic flux


ii.; = f!oFim/8o = V 2" lnif!oIikwiwi/8onp

rotates at an electrical angular velocity WI = pQ 1 [see
Eq . (25-3)1 in the positive direction (this is, from phase A
towards phase B).
The fundamental mmf of the secondary winding

Fzm = V2 InzI2kwzw2/np
or the fundamental component of the air gap magnetic flux

rotates relative to the rotor at an electrical angular velocity W2 equal to the angular frequency of the current in the
rotor winding, and does so likewise in the positive direc-


ell. 29 Electromagnetic Torque

tion, that is, from phase a to phase b. (In Fig. 29-'1, the velocity of the mmf relative to the rotor, w2 , is shown relative to
the rotor). To find the angular velocity of the secondary
mmf F _ relative t o the stator, w'2 , it should be recalled that

F ig . 29-1 Relative position of the fund amental mrnls and flux lin
. '
kages in the primary and secondary wi~dings

in the model the rotorrotatesatan electrical angular velocity W = Qp in the positive' direction, so this velocity must
be added t o that of F z relative to the rot or
w~ =

+ oi

(In Fig . 29-'1, W z is shown likewise relative to the rotor).

Since the condition for t he velocity of rotor cur rent is
satisfied, we may wri te
-' w ~ =

WI -

+ W = ' WI

= QI l'

To sum up, unidirectional energy conversion can be

performed only if the rotor mmf rotates with the sameelectrical angular velocity WI = QIP in t he model and with the
same mechanical angula r velo city Q I in the prototype machi23 -0169


Part Two. Energy Conversion by Electrical Machin es

ne as the stator mmf does. (In Fig. 29-1, cu ~ = CUI is shown

relat ive to t he sta tor .)

Th e conv erted energy and t he period-aver aged elec tromagnetic torque depend , as will be shown la te r, on the ele ctrical angle a12 betw een t he ax es of t he stator and ro t or
fields stationary relative to each other. This angle is connecte d to the mechanical angle 1'12 betw een t he sam e axes in
the prototype machine by a known relation, a 12 = PY1 2'
The positive direction for a 1 2 (or 1'12) and for the torque
act in g on the roto r is counterclockw ise from the ax is of the
rotor field .
Because under steady-state conditions a 1 2 is constant,
the t orque will likewise remain cons t ant over a rev olu tion,
and the period-averaged electromagnetic torque can be fonn d
by Eq. (18-6)
T em

= aw/ay

lin = constant

for the arbitrary relative position of the sta tor and rotor
shown in Fig. 29-1. (The angle a between the axes of the
main rotor and stator ph ases is arbitrary .)
In order to find the electromagnetic torque, we should
first determine the energy of the air gap magnetic fiel d as
a function of B l m, B 2m, and a 12. Th e energy of the magnetic
field in an elementary volume dV of the air gap is

dW = (Bg/2flo) dV
where B o = B om cos (pcp) = magnetic flux density in t he
elementary volume dV = l o<'3 oRd cp
B om = V Brm + B~m + 2B lmB2m cos a 1 2 = peak flu x
densit y of the resultant air gap fi eld
cp = angle defining t he position of the element ary
volume relat ive t o t he resultan t fi eld
R = mean air gap radius
The energy of the air gap magnetic fi eld is found by takin g
t he integral over the volume, V = 2nRlo8o. It is



r (B~/2~to)


dV =

) (B5m/2flo) loooR cos- (pcp) drp

2B 1m B 2m COS a 12)
1m + 2m +

= nR/p is the pole pitch.


eh. 29 Electromagneti c Torque

. . Now we turn the rotor through a small angle aI', deeming

the current constant, and find dW/dy. As will be recalled,
the mechanical angle I' (or the corresponding electrical
angle a = yp) is the angle, say, between the axis of the
stator phase A and the rotor phase a (as reckoned from the
stator phase A in the positive direction , that is, counterclockwise) . As the rotor turns through a small angle dy = da/p
in the positive direction (with t he phase currents held constant) the rotor mmf and field move along with the rotor,
whereas the stator mmf and field remain stationary (see
Fig. 29-1). In the process, the angle 1'12 = a 12 /p between the
rotor and stator mmfs decreases in the same proportion as
the angle I' increases. (We have assumed that t he angle 1'12 is
reckoned from a rotor phase to a stator phase, that is, in
the revers e direction from that for the angle I' or cc .) As
a consequence, the changes in I' and 1'12 only differ in sign
dy = da/p = - dl'12 = - da12/P
dee = - da l2
H ence , it is legitimate t o write the derivative as follows:
T em = -d- = - P -d- =

p2-r:ool r,B1mB 2m

0: 1 2

B lm

= const,

B 2m

SIn a 12


= const

Expressing the magnetic flux densities in terms of currents

and recalling Eq . (28-3), we get

T em =


I 12m
. a12

where L m is the maximum mutual inductance between the

stator and rotor phases as defined by Eq . (28-3) .
The elec tromagnetic torque acting on the rotor is positive
(that is, is directed counterclockwise) when 0 < a12 < rr,
and negative when n < a l2 < 2n (or 0 > a1 2 > - rr). On
expressing B 2 in terms of 1 2 and noting that


is the peak flux linkage of the stator field with a rotor phase,
we may write the electromagnetic torque in t erms of current
and flux linkage as
2P I nr
T em = mV2
2 T 21m SIn a12


Part Two. Energy Conversion by Electrical Machines

or, in complex notation ,

I [nr 1~*]
T em= m2P
l fz m T 21 2
where ~' is the complex conjugate of the secondary (rotor)
Using the above equations, it is an easy matter t o show
tha t the interaction of 1 2 with its own tield or flux linkage
qr 22 produces no electromagne tic torque . To demonstrate,
by Eq. (29-2) , this to rque is zero:
z 12HI 22m SIn
CG 22 =


where qr 22m = 2 B 2 m L loW 2kw 2/ :rr, is the peak flux linkage of

the useful self-field with the rotor turns, and CG 2 2 = 0 is the
angle between qr 22m and 1 2 ,
Now we are in a position t o express the electromagnetic
torque in terms of the total flux linkage with a given winding,
If 20m' that is, in terms of the flu x linkage produced by both
the external field , If 21m' and the self-fi el d, qr 22m' For t his
purpose, we add to the right-hand side of Eq. (29-2) the zero
torque associated with the self-flux linkage
T e m --

m 2P I

= V~


TU m


Im [(W21m

2 - -

m 2P I

[lIf 22m ~
1* ]

+ 1Jf22m ) 11]

= m2 Pz I m [ill
T20m 2

-_ m2P


20m 2 sin CG20


ijr 22m -qr 21m = peak t ota l flux linkage 'of

the main field with the ro tor winding
CX 20 = angle between 1; (or B2 ) and "If 20 m (see
Fig. 29-1)
A torque, equal in magnitude but opposite in direction,
is also acting on the stator. It can be found by Eq . (29-1)
or Eq. (29-2), recalling that the t or que at the stator is
deemed positive when it is acting clockwise (that is, against
the sense of rotation). Alternatively, the electromagnetic
torque at the stator may be expressed in terms of sta tor


"if 20 m =


Ch. 29 Electroma gne tic Torque

qu antit ies. To t hi s end we write B I in E q. (29-1) in terms of

II an d recall t hat
Th en ,

t ern = .
' a1"
, r - 1If12m I 1 SIn


v 2

, I m ['If*10 1-]
1/ 2 -


where 1Jf 12m is the peak flu x li nkage of the rotor field with
a stator phase.
Since the in teraction of II with the self-field or self-flux
pro duces no electromag net ic torque, i. e.

it is an easy matter t o express t he electromagnet ic t orque at

t he stator in terms of t he to t al flux linkage of the useful fi eld
with a stator phase

T em


- r = - r10m

1/ 2

I 1 S I. n

a10 =

,rV 2

Im ['If*10 m I~1]


1jJ12m if11m = peak t otal flux li nk age of

t he useful field with a st ator phase
a I O = angle between -Plom (which is in phase with
-p 20m) and L (as reckoned counterclo ckwise
from flu x linkage to wards cur rent).
Equations (29-4) an d (29-5) are equally applicable t o
a polyphase ro tor winding and a d.c.-en ergi zed single-phase
rot or winding suc h as used in synchronous machi nes. '
Equations (29-3) and (29-5) where the elect roma gnet ic
torque is expre ssed in terms of 1ff 20 m or 0/ 10m also hold for
sa turable ma chines with nonlinear cores. Deri ving t hem
subject t o t he remarks ma de in Sec. 18-2, it will be seen that
to fin d t he elect romagne t ic t orque in such a case it will
suffice t o sub st itute into Eq. (29-3) or (29-5) the peak values
of the fun dam ent al flux li nk ages found with allowa nce for
the nonlinearity of the ma gneti c circuit,

wher e



Part Two. Energy Conversion by Electrical Machines



The Electromagnetic Torque Expressed

in Terms of Electromagnetic Forces

In the previous section, we have found the electromagnetic

torque from the law of conservation of energy. It can be
determined in other ways as well. For example, we could
combine the torques due to the electromagnetic forces
which arise when a rotating magnetic field interacts with
the elementary currents and elementary surfaces of the
magnetized cores. We could then have obtained a more
detailed picture about the distribution of electromagnetic
forces throughout the active parts, the flows of energy converted by a given machine, and their directions. Unfortunately,
the mathematics involved would be prohibitive out of an y
proportion. Therefore, if we are only interested in the main
electromagnetic torque associated with the fundamental
mmf and airgap magnetic flux density, it is convenient to
use the concept of the surface current which replaces the
currents in the core slots.
(i) Surface Current and Its Fourier Expansion
We obtain the surface current on replacing the toothed core
by a smooth one and spreading each slot currents i s /n over the
core surface as a thin sheet with a linear density given by
Ash = ish/b s
The replacement of slot currents by a surface current is
illustrated in Fig. 29-2 which repeats the winding and current patterns shown in Fig. 25-8 for i'.4. = V:2 I a and in
= i c = - 112 I a/2. Shown below the sectional view of
a slot layer in a toothed core is an equivalent smooth core.
The air gap is enlarged k 6 times, and the currents are
shown spread outside the slots but within the slot boundaries as thin sheets of density ASh.
The slot current ish is the sum of alternating currents in
the conductors laid out in the kth slot. For example, the
current in slot 2 of Fig. 29-2 enclosing the forward conductors
of phase A and the reverse conductors of phase C is
i S2

iAlOC -

iclO c

At the instant of time shown in Fig. 29-2, when

iA = i A m a x =

-V 21a

i p=i 9 = -1I2I~/2


Gh. 29 Electromagn etic Torque

the current in the second slot will be

i S2 =

V2" Iawc + V2" I awc/2 =


V2" I awe

The current ish and the corresponding surface current density

for the kth slot, A sh = ish/b s ' are t aken t o be nega tive if the
cur rent is shown flowing "inwards" (away from -the read er),
Ax i s A


PI. or


Fig . 29-2 Surface current densi ty and mmfs of a pol yph ase winding
= 3, q = 4, i A = V2I a , in = i c = - lffl a/2)


that is , with the Z-axis. Th er efore in , sa y, slot 9 the current

at the instant of t im e in question is negative
\ iso= i nwc+i nwc= -l/ZIa we

Betwe en slots, the surface current density is zero . Th e

cycle of change in the surfa ce curren t density is the same as
for slot currents. R ound the peripher y of the air gap , it
changes p t imes (wher e p is the numb er of pole pairs) . A~


Part Two. Ene rgy Conversion by Electrical Machines

slot cur rents va ry , the wa veform of surface current densi t y

also varies in .a continuous fashion. On expressing slot currents as functions of time and expanding the spatial distribution of surface current density in to a Fourier series, we
could find th e fundamental surface current density with 2p
poles round the periphery an d with peak va lue A Im, as shown
in Fig. 29-2.
Unfortunately, the above procedure is too time- and effor t
consuming. ' A far simpler appro ach is to expr ess thes urface

Phase axls A

Fig. 29-3 Relationship between the tangential field intensity and
surface curr ent density (axial component) in polar coordinates

current density in terms of the mmf or of its harmonic COIlJponents. sPrior t o that , it is neces sary t o find t he tangential
component,H'\', of the airgap magnetic in tensity on the
smoo th cor e surface where t he surface current is distribut ed
as an infinite thi n shee t of thickness 11 = 0 an d of a linear
density A (Fig. 29-3) . In polar coordinates , the surface
curr ent density A , and al so the slo t currents, are directed
along th e Z-axis, and A = A z - Let us find the current l1i
for a surface element of length R 111' .
, l1i = A x . R 111'
Enclose t h is curren t by a rec tangular loop labelled 1-2-3-4
an d having a r adi al dimension h and a tangential dimension
R (111') . Applying Ampere's circuital law t o the circulation
Of t he ma gnetic in tensity round the loop 1-2-[/-4 where h -+ 9


Ch. 29 El ectromagnetic Torque

and noting that the magn etic intensity on sid e 1-4 lying
within the core of an infinite permeability is zero, we get

n, dl = H1'R!1y = !1 i

On passing to the limit with !1y --+ 0, we ohtain

H l' = H t = lim (!1i/R !11') = A

This is the expression for H 1" t he magnetic intensity on the
surface of the inner core . Applying the same procedure , we
can obtain an expression for the magnetic intensity on t he
surface of the outer core, H'I' = - A.
To sum up , the tangential component of the magn etic
intensity, H 'I' = H t, on the surface of the core with ~ta = 00
is equal in absolute value to the current density on that
surface , I A !.
Proceeding from Eq . (23-9) and setting dx = Rd'\' in polar
coordinates, we can write the tangential magnetic intensity
as a derivative of the scalar potential on the core surface
H 'I' = -

drp/R d'\'

Finally, using Eq . (24-5) . and the accompanying relations

between the mmf and the distrihution of rp on the excited
core surface (F = qJ when the inner core is excited , and
F = - qJ when the outer core is excited), we can express the
surface current densit y on the inner or outer core as a derivative of the mmf

dF/R dy


This relation can be applied not only t o the surface current

density as a whole , bu t t o its harmonics as well, so that for
each harmonic mmf there will be a surface current density
harmonic of its own.
Now we set out to find the fundamental surface current
density when a pol yphase winding car ries PPS curren ts of
frequency CD . The fundamental mmf is a rotating wave described by Eq. (25-10). Proceeding from Eqs. (29-6) an d (25-10) ,
we have
A = -

dF/R d'\'

A I m cos (CDt -I- n/2 -



In terms of the peak value of t he fundamental mmf, the

peak value of the fundamental surface current density; A ln~ !


Part Two. Energy Conversion by Electrical Machines

can be written
If we use Eq . (25-16) for F i m, then Aim can be expressed
in t erms of the sum of rms slot currents , 2wc I a = I s as

AIm =

11 2" Aokpkd


where A o = 2wcIait z = Is lt z is the line load current found

from the sum of slot curren ts .

Fig . 29-4 Fundamental compone nt of surf ace current density (in a

two-pole mod el)

At a point at an angle a to the origin , t he surface curren t

density can be presented as the projection of the compl ex
surfa ce current densi t y

v.; (nIT) =

Ai m exp [j (wt

+ n /2) ]

all the axis at an angle a, t hat is,

A = Re [A i m exp (-ja) ]
Using Eqs . (29-7) and (29-9), it is an easy matter to plo t
the fundamental component of surface current density on
the same scal e as th e slot current densities. This plo t is
shown ill Figs. 29-2 and 29-4. Figure 29-.(' al so shows the
surface cur ren t den si t y phasor Aim.


Ch. 29 Electromagnetic Torque

From the foregoing, we may conclude that the fund am ental

component of linear surface current den si ty is a rotating
wave with period 2'1: and amplitude A ]m, t ra vell ing at the
same mechanical angular velo city Q (or an elect rical angular velocity CD in the model) as the fundamental mmf. Irrespective of the direction in which the mmf is ro tating, the

Fig . 29-5 P attern of th e ma gnetic field set up by currents in the

inner core of a machine

surface current densi t y wave always makes an angle of n/2

with the mmf wave or is displaced through r /2 counterclockwise.
The above rel ation between t he tangential magn etic in t ensity and t he surface current density, Il , = H 1, = + A , also
holds for their fundamental component s. Therefore , t he air
gap magnetic field set up by t he slo t cur ren ts (or by the
fundamental component of the equivalent surface current
density) always has the tangential as well as the radial
com ponent . Th e radial component of t he magnetic fl ux
densi t y can be written in terms of t he mmf using Eq. (25-18) as

n, =

lkoF/k 1i6


The tangentia l magnetic flux densit y on the surface of t he

excited core may be written directly in t erm s of the surface
current as
B t = B i' = ~t oH l' = + ~toA
Sin ce no tangenti al magneLic flux dens ity exists on the
surf ace of the unexcited core , t he air gap fi eld set up by the
funda ment al component has t he pattern shown in Fig. 29-5,


Part Two. Energy Conversion by Electri cal Machines

It should be noted, t hough, that the figure shows the field

for a very large relativo gap , 6IL , when the radia l component is comp arable with the tangential component; compare
Eqs . (29-10) an d (29'-11). For the small values of oIL usu all y
encountered in practice , the radial component is sub stantially larger t han the tangential com ponent , so the field is
essentially a radial one. If, however, we neglected the
tangential components, H I" due to t he surfa ce current of

Fig. 29-6 Relative position of rotating waves of radial magnetic

flux dens ity B ; and surface current density A

density A, we would be unabl e t o get proper insigh t into the

generation of electromagnetic forces and transfer of energy
across the air gap when an electrica l machine is running.

(ii ) Elec tromagnetic Torque

Let us find the electromagnetic torque acting on the rotor
rot at ing at a mechanica l angular velocity Q (or at an electrical angula r velocity ro in the model of Fig. 29-6) . Sup pose
that the poly phase rotor wind ing carries a current /2 with
angular frequency w2 , wh ich gives rise to t he fundamental
surface curren t densi t y wave of peak va lue A 2 m The mech anical angular velocity of the surface current density wave
relative to t he rotor is proportional to the frequency of the

Gh. 29 Electromagnetic Torque



Q 2 = w 21p
The electrical angular velocit y of th is wave relative t o the
rotor in the model is t he same as that of t he cur rent and is
equal to W 2
Relative to the st at ionary frame of ref eren ce, the surface
current wave ro tates at a mechanical angular velocity
QI = Q + Q2 ' The electrical ang ular velocity of the wave
relative to t he stator in the model, WI = W + W 2, in the
case of uni directional energy conv ersion (see Sec . 29-1) is
always the same as the angular frequency WI of currents in
the pol yphase stator winding.
Therefore, the fundamen t al
stator mmf rotates at a mechanical angular velocity QI
= wIlp and produces, toge ther
with the fundamental component of the rotor surface current rotating at t he same mecha nical angular velocity, t he
useful rotating magnetic field
in whi ch the radial magnetic
flux density has a peak value
given by B I (O)m = B om . Let
the angle between B o and A 2
(or between Bom and A2m ) be
Fig . 29-7 Electromagnetic fordenoted by P02 and counted
ces acting on an element of from A~2m towards jj om'
surface current
The electromagnetic t or que
can be defined as t he sum of
the torques developed by the electromagnetic forces dN
acting on elementary surface currents di = A 2 (R dy).
Assume t hat each elementary current extends along the
machine axis for a distance equal to a unit of length and is
lying in a magnetic field in which the radial magnetic
flux density is B r Then t he elec tromagnetic force that is
acting on that elementary current in a tangential direction
ma y be written as
dN = B ; di = B r A 2R dy
The direction of dN can be found by t he left-hand rule.
An elementary surface current at an angle l' to the origin
and the force acting on it ar e sho wn in Fig. 29-7.

Part Two. Energy Conversion

by Electrical


The electromagnetic torque acting on the surface current

and transmitted to the shaft by the mechanically stressed
rotor core can be found by adding together the torques produced by the elementary forces
dT em = R dN
over the entire rotor surface facing the air gap (from '\,
2:n: along the effective core length lo)

0 to

'\, =



T em = lo 1 RdN = loRz1 BrAzd,\,

Using Eq. (25-18) for a rotating magnetic flux densit.y wave
B; = B om cos (WIt - a)

and Eq. (29-7) for a rotating surface current wave

A z = A zm cos (WIt - a -


where a = py, and on setting for simplicity Wit= 0, because

the integration yields the same result for any instant of
time, we obtain

Tern = :n:RzloBomA2m cos P02 = -n- loBomAzm cos Poz


Noting that the peak value of the linear surface current density can be expressed in terms of I z [see Eqs. (29-8) and

A zm = :n:Fzm/-t =

ev 2 m2/-t) (Izwzkwz/


and the peak value of the magnetic flux density defined by

Eq. (27-2) can be expressed in terms of magnetic flux
B om = :n:cI>m/2'tlo
the electromagnetic torque can be written as a function of
the magnetic flux and winding current

Tern = (pmz/V 2) (wzkwzcI>m) I z cos Poz


If we recall that the peak value of the total flux linkage

with a phase given by Eq. (27-14) is

"IfZOm = w2k wzcI>m

and replace Poz by sin a oz = sin (n/2 + Poz) = cos P02

[see Fig. 29-6 and Eq. (29-7)] 1 the expression for the torque


Ch. 20 Ele ctromagne tic Torque

will be analogous with Eq. (29-3):

Tern = (pmz/V

2) 'zom1z sin a oz

where a 0 2 is the electrical angle (see Fig. 29-6) between the

flux linkage phasor"lf 20m and the current phasor 1 2 (or
the mmf phasor F 2)'
As already noted in Sec. 29-1, the torque equation,
Eq. (29-3), also holds for saturated machines with a nonlinear
magnetic circuit. A further proof of that statement is the
fact that it is analogous to the equation derived here for
a saturated machine.


Electromagnetic Force Distribution

in a Wound Slot

In finding the electromagnetic torque acting on the rotor,

we replaced the toothed rotor whose slots carried certain
currents is by a smooth core, mo ved the currents is to the
surface, and distributed them in the slot regions as an infini te thin sheet of linear density A s = i s/ b s (see Fig. 24-2).
The replacement of a toothed by a smooth core will leave
unchanged the tangential electromagnet ic force N acting
on a slot, if the mean air gap magnetic flux density due to
the external magnetic field is as found with all owance for
the saliency of the stator by the equation
Bo,mean = fJ-o F/{) k 01k oz

where F

k 01 =
k oz =

() =

external mmf produced by the stator currents at

t he axis of the slot in question
stator air gap factor
rotor air gap factor
radial gap length.

The total tangential electromagnetic force per unit length

of air gap can be found as the sum of the for ces

B o,mean Asdx

applied t o the elementary surface currents

+bs/ Z

-b s/2

s; rneanAs dz = s; meanAsbs = B o meani s


Part Two. Energy Conversion by Electrical Machines

Let us now see how this force is distributed in the slot

region. The resultant magnetic field in the slot region may
be visualized as the sum of the external field (Fig. 29-8a)
and the self-field set up by the slot current (Fig. 29-8b). To
determine the tangential forces acting on the slot sides and
the slot current, it will suffice to find the respective flux
densities in the slot sides and at the centro of a currentcarrying conductor. As has been found for real slots and an
infinite core permeability (~~I'C = (0) , nearly all field lines
entering a slot terminate in the slot sides and only a small
fraction of the lines reaches the current-carrying conductor.
If the flux per unit slo t length (see Fig. 29-8a) is
+ bs/ Z

cD =

J' B I1(y=o) d l : = B o, meanbs


then the flux passing through the slot section at the level of
the current-carrying conductor (y = hi) will he


~ BY(Y=hj)dx=Bsb s
- "s/~

where c ranges from 0.002 to 0.001 .

. Because the magnetic field is symmetrical and continuous,
the flux entering the slot side

cD o = (cIJ


differs but little from cD/2.

On moving away from the air gap, the external magnetic
flux density on the slot sides, B 01 and B 02' rapidly diminishes so that level with the conductor top (y= h) it is zero
very nearly. Therefore , the external flux enters the slot
sides within the depth 12

cPo = cI) (1-c)/2 = ~ B 0 1 dy = ~ Bozdy


From a comparison of the expressions for c~ and ccD the

external flux density in the conduct or region can be written

cB 0 .



Ch. 29 Electromagnetic Torque



- BiZ

////////////////// / /!/I/ /4


Fig. 29-8 Distri bution of elect romag netic forces in the reg ion of a
wound slot:
(a) external magne ti c field due to currents on the ot her core, I s = 0,
B /i,mean=(= O; (b) magn etic field due to slot cur ren t , B/i,m ean=O , 1 s=
= 0; (c) electromagnetic forces in and aro und a slot, I s =(= O, B/i,mean=O

24 -016 9


Part Two. Energy Conversion by Electrical Machines

The pattern of the field set up ;by the slot current and the
distribution of the magnetic flux density in the slot sides are
shown in Fig. 29-8b.
Within the larger part of the space above th e conductor
(0 < y < h), the magnetic flux density in the slot sides is
nearly constant. Applying Ampere's circuital law to the
loop enclosing the current and coincident with a field line
over ' the slot width , the respective magnetic flux ' "densit y
can aproximately be written as

B il

B i2 = ftoic/bs

Now we are in .a position to find the tangential electromagnetic forces asso ciated with the resultant magnetic
field in the slot region (see Fig. 29-8e). The curren t-carr ying
conductor is acted upon by a tangential electromagnetic
force given by


Bci c

eB a,mean i c


which is a small fraction of the total force

N = B a,meanic

acting in the wound-slot region.

The greater part of the total electromagnetic force is
applied to the slot sides as a magnetic pull. Let us
first find the specific magnetic pull, T (in pascals),
th at is the force per unit area of a ferromagne tic surface
(ftr ,Fe = 00) in a magnetic fi eld . It can be expressed in terms
of the normal component of magnetic flux density, B n ,
whi ch acts at the surf ace of a ferromagnetic bod y in the same
direction as the total magnetic flux density, and in terms of
the permeability fto of the nonmagnetic medium surrounding
that body:
T = B212fto
The specific magnetic pull vector is always aligned with the normal ;; to the surface of the ferromagnetic
body, directed towards the medium having the lower permeability
T = ;; (B2/2fto)
The specific magnetic pulls T 1 an d T 2 acting on the
slot sides are normal to the sides. The dis tribution of


Cit. 29 Electromagnetic Torque

T I an d T z

T z = B;/2~Lo

T I = Bi / 2 f-lo,

over the slot depth is a funct.ion of


B Ol -t- Bi,

on th e left-hand sid e, and of

B ; = B 02


on the right-hand side . Because they are unbalanced only

over the int erval 0 < y < h, the resultant tangential force

N ss


N ssz

applied to the slot si des can be found b y combining the

ele m en tar y forces on each side within the limits specified:

N ss=NsSI-Nss2=



T 2dy =


J(Tt-T z)dy

Noting that
we get


T 2 = (2/ ~Lo) BolB il

In taking the integral , i t should be re called that

B i l ~ f-loic /bs

cons tan t

for 0 < y < h (see above). So, on expressing the flux

through a slo t side




in terms of the external flux, cD

= bsB 0 , mean'

N ss=

we get


J(Tt -T z) cly= (2/f-lo) Bi J B



= (B ii/f-lO) (1- c) <D = ('1- c) icBo ,mean = (1- c) N
The total tangential force N acting on the wound-slot region
is th e sum of N ss' the force applied to the slot sides, a nd
2 1~ *


Part Two. En ergy Conversion by Electrical Machines

N e , the force applied to the current-carrying conductor,

N = N ss

+Ne =

(1 -

e)B Il,mean ie

+ eB Il,mea nie

= B Il,meanie
This for ce is equal t o the for ce ac ting on ie ' the slo t curran t
shifted to the core surface. As we have learned , howe ver, t he
greater proportion of this force acts on the slo t sides r aLh er
th an on t he conductor in a slo t (e = 0 .00 1 to 0. 002).
Our reasoning has been based on cer tain simplifying
assumptions as regards the distribu tion of the ex ternal flu x
density in t he air gap an d of t he flux densi t y du e to t he
slot current in the slot sides. However, th e rigorous appr oach
would lead t o t he sa m e solution - a fac t of impor tant practical significance. Because electr omagnet ic forces l argel y
act on the slot sides (or the core t eeth), the conductor insula tion m ay be designed as me ch anically s tr ong as m ay be necessary t o t ransfer N'; = eN which is a very small quan tity.
To sum up, owing to the shielding action of teeth on
a toothed core, the exter n al fi eld in the region taken up by
current-carryin g con ductors is substanti ally reduced , and
the requirements for the mechanical st re ng th of insulation
m ay be less str ingen t .


Energy Conversion
by a Rotating Magnetic Field

30- 1

Electromagnetic, Electric
and Magnetic Power

The elect r om agneti c power en taring a 1'0 tor surface element dS = '1 X Rely from the ail' ga p is a fu n cti on of the
powe r dev elop ed by an eleme n t of tor que as an ele me nt of
current rotates at a m echanical angular vel ocity Q 1 (see
Sec. 29-2)
Le t us writ e t he me ch anical angul ar ve locity of a surface
current el em en t as t he sum of Q , the rotational angular
veloci t y due to the rotat ion of t he 1'0101', an d Q 2' the ang ular
velo ci t y of t he curre nt sheet rel a ti ve to the rotor body,


Ch. 30 Energy Conversion by Rotating Field

associa ted with period Ic variations in the curren ts of the

rotor winding:
Q] = Q


Then the electromagnetic power entering a rotor surface

element can he written as the sum of the mechanical power

dP m = Q dT
transferred via this surface element to the shaft, and the
electrical power
entering the sur fac e current elemen t or the winding to
which it is equivalent :
dP e m = Q 1 dT


QdT + Q 2 dT
dP m

+ dP e (30-2)

power flow per unit area
(power flux density) and the
direction of this flow may
he defined in terms of the

radial component, 'fl' e m , of

the Poynting vector [24]


Fig . 30-t Flux density: elect ro magnetic power (P em), elect r ic power
(P e)' and mechanical power ( Um )
(on th e su rfa ce of th e rotor core)

= E em X H I'
= q zq i,E emH i '
= - q,.E emH y

which is seen (Fig. 30-1) to

he the product of the axial
com ponent of the electric in l ensi ty vector, E e rn = q zE em, and
the tangential com ponen t of the magnetic in tensity vector,
H I' = Cl I'H I" H er e , Cl i " (I " an d Cl ,. are the unit vectors along
[he respective axes of a cylindrical coordina te system .
If we recall that th e tangential magn etic intensity on the
rotor surface is equal to the surface current density (see
Sec. 29-2)
HI' = A 2
ancl al so tlia L in defining the to tal energ y en tering a surface
element we should take into account the electric intensity


Part Two. Energy Conver sion by Electri cal Machines

found from the total lineal' velo city of a surface current wave
"1 = fJ VUI = q vRQlj

that is ,

E em

B, X " I

fJr X q vBrUI = q zBrRQ I

th en the radi al componen t of the Poynting vector may be


~ em =


= -

,fT' em

+ fJ r ;P em

E emH v


= - B r R QIA 2

Naturall y, the electromagnetic power ent ering a r otor

surface element , '1 X R dy, and expressed in terms of th e
radial component of t he Po ynting vector

elPem - =. ~ emR dl' = - BrRQ IA 2 (R ell') = - Q I d T

is the sam e as given by Eq . (30-1). The " _ " sign implies
that the power flow is directed inside t he rotor, that is,
oppos ite to the radial uni t vect or qr ' In a similar man ner,
we can represent the electric power entering a surface curre n t
eleme n t. In this case, however, the elect r ic intensi ty must be
dedu ced from the' velocity of the surf ace current wav e,
solel y related to vari ations in the winding current s:

dP e = ff'e R dl' = --

fJ vU2 = q VR Q 2
(B rRQ 2 ) A 2 R dl'

= -

Q 2d T



where ff' e = E e X H v = qr ff' e, an d f? e = - E cH v

= - B r R Q 2 Here, the " _ " sign likewi se indicates tho
directi on of the power fl ow (see Fig. 30-1).
By th e same token , the mechanical power
f1P m = Q dT
ma y be expressed in te rms of th e radial componen t of th e
Umo v vector , D , defined as the mechanical power flux en tering a surface element of a mechanically straine d body .
Th e radial component of th e Umo v vec t or fo r a rot ating
body is determined asthe product of the taug en tial mecha nical stress T v by the ' ,t angent ial linear di splacemen t velocity of a surface element , /I.., = u. = RQ



A-aBr R Q



Ch, 30 Energy Conversion by Rotating Field

The tangenti al mechanical stress 't v arises on the outer

surface of the ro tor because it carr ies surface curren t elements
and the tangential forces d N act ing on t hem
't v =

dN/R d'V =

(qzA2R dl') x qrBr

R dl'

CI v't v


wh ere 't v = AzB r [N m "].

The mechanical power entering a rotor surface element

is the same as given by Eq. (30-2) wh ere it is expressed in

terms of the electromagnetic torque.
By taking an integral over the rotor surface , we can find
the electr oma gnet ic and electri c powers t ha t en tel' the
rotor in th e process of energ y conversion:

P ern = l o


dP ern =

u.i, j


Pe =lo

; =



= Q 1T


j dP e =Q j dT =Q T "
2l 0



j dP m = Ql o j"


dT = QT

Obviousl y, as follows from Eq. (30-2),


; + Pm

The electroma gne t ic power , P em ' is the total power tra nsferred by the ro tating fi eld t o the rotor (Fig. 30-2). Some of
this power , P e- is dissipated as heat in t he ro tor windi ng or
in t he line connec ted to that winding. This can be proved
by re-arranging Eq. (30-8) wi th t he aid of E qs. (29-3) and
(27-'1 5) and also rec alling tha t ~02 = ao z - n /2 is, at the
same time, the angle hetween I z and E z in the rotor winding. Therefore ,

Part Two . Energy Conversion by Electrical Machines

where pQ 2 = (02 is the electrical angular frequency of E 2

induced in the rotor winding by the rotating fie ld .

Fig. 30-2 El ectromechani cal energ y con versi on by a machine (on

t he left) and in its model (on the ri ght)

Finally, the remainder of input power is converted to

mechanical power
Pm = QT
transmit ted by a mechanically loaded shaft (as the power
flow defined by the Umov vector) .

* 30-2

Energy Conversion in an Electrical

Machine and Its Model

As we have seen in the previous chapters , t he two-pole model

of an electrical machine is a convenient and instructive tool
with which to study what goes on in the machine itself.
This is also true of th e even ts in vol ved in energy conversion .
In the model shown in Fig. 30-2, we retain the prototype's
pole pitch T, fundamental amplitudes of magnetic flux

Ch, 30 En ergy Conversion by Rotating Field


densi ty and surface current waves , B 0 111 and A z, and also

phase displacement between these waves in fractions of a pol e
pitch. Therefore , in the mod el the wa ves tr av el thr ough
an angle ~oz which is p times as great as the angle between
these waves, ~0 2Ip, in the machine itself . (On th e left. of
Fig. 30-2 , t he m achine's field has four pol es.)
Also, we retain in the mod el th e proto type's linear peri pheral velocities of the waves and of the cor e element , U 1 , U z
and U, on the outer surfa ce of th e rotor. This is don e because
in the model the r adius of this surface is 1lp of its value in
the prototype, whereas the angu lar velocities are p times as
U 1 = (R ip) QIP = Q1R
liz = (R ip) Qzp = QzR
U = (R ip) Qp = QR

The t angential electromagne tic force acting on the surface

current in the model's rotor is the same force as operates on
the surface curren t over a pol e pitch in t he pro to t ype , that
is N ip, which is 'lip as large as the tot al tangential force
(see above)

N = i,

J dN

The electromagnetic t or que in the model is 'lIpz of its

magnitude in the prototype, because it is given by th e
product of th e tangen t ial for ce in the model , N ip , by t he
radius of its ro to r , R ip :
(N ip) (R ip) = N R lp? = I'lp ?
Ho wever , the powers entering t he r ot or of t he model do
not differ from t he respective powers exi sting: over a pol e
pitch in the prototype machine. This can be demonstrated
by applying Eq. (30-8) 't o the model:
(Q1P) (T lpZ) = Q1T lp = P emlp
(Qzp) (T lp Z) = QzT lp = P elp
(Qp) (T lpZ) = QTlp = Pmlp
The same relations can be obtained , if we re call that at
the similar poi nts over a pole pitch in the pro to type and
in its model (that is, the points wh ich are a t angles '\, and


Part Two. Energy Conversion by Electrical Machines

a= P"Y, respectively) , the radial magnetic flux densi ty B n

the linear surface current density A , and all the velocities

are respectively the same. Accordingly, P em, P e and Pm'
definable in terms of the radial components of the Po yn ting
ane! Umov vectors, Eqs. (30-3) through (30-5), are .likewisa
the same at th e similar poin ts in t he ma chine and its model.
As is seen , the model is convenient not only in calculating
th e air gap fiel d, mmf, emf, and flux linkage, and in plotting

Fig. 30-3 Electrical ma chine an d an equiva lent system of p twopole model s

th eir respective ph asor an d vector diagrams wh ere these

qu antities app ear as phasors and/or vectors, but also in
ana ly zing power fl ows . In doi ng so, it is import an t to hear
in mind tha t the powers in the model are 1/p of their magnitu des in t he pro tot yp e ma chi ne. Therefore , as regard s .energy
conv ersion, a 2p-pole machine maybe replaced by a system
of p models ha ving a comm on shaft an d connecte d to the
same lines and ro t ating at a speed multiplied p t imes*.
Obviously, such a model system will handle the same powers.
A four-pol e ma chine and an equi vale nt syst em consisting of
two models are shown in Fi g. 30-3. The pro totype is a syn-

* In stead of combining p models wi th th e same core length l6 as

in th e prototype ma chine, we ma y use one two-pole machine in whi ch
th e core length is p times as large'. .

Ch. 31 Energy Conversion Losses


chronous motor (see Part 5) in which the rotor winding is

energized with direct current (W2 = 0) and the me ch an ical
angular velocity of the field, Q l ' is the same as the mechani cal
angular veloc ity of the rotor , Q . Therefore, all of t he electromagnetic power, P em = Q1T, applied to the rotor is converte d to mec han ica l power, P m = QT, and the elec tric
power, P e , entering the rotor winding is zer o
F ; = Q2T = 0
The directions uf the power fluxes , electromagnetic forces
and torques hold for the motor mod e of operation . The
external to rque acting on the motor shaft is denote d by Text .


Energy Conversion Losses

and Eff iciency


Introductory Notes

As has been shown in Cha p . 2'1, for electric energy , P " to

be conver ted into mechan ica l energy , P m' or back by a rotating electr ical machine , the following conditions must be
(i) The rotor whose shaft transmits mechanical energy
must be rotating continuo usly .
(ii ) The windings must carry currents whose frequencies
are related to one another and to the mechanical angular
velocity of the r otor in a certa in defin ite manner.
(ii i) The m agnetic fl uxes li nki ng the wind ings that are
responsib le for energy conversion must vary per iodi cally .
As a consequence, some of t he energy han dl ed by an electrical machine is inevit abl y dissipat ed owing to fr iction
between t he rotating parts; t his is what is known as mechanical losses. Another fracti on of the t otal energy is lost as
currents t raverse the wind ing conductors; this is electrical
losses. Still another fraction of the total energy is lost as
the cores underg o cyclic m agn eti zati on ; t his is magnetic
losses .
All kinds of losses are customarily expressed in terms of the
equ ivalent t hermal energy dissipated pel' unit time or the
time rate of energy loss, "x'P . In our subsequent dis cussion, it
will be collectively called the power losses.


Part Two. En ergy Conversion by Ele ctrical Machines

From th e law of conserv ation of energ y , it follows that

the useful output power from a machine is alwa ys smaller
than its input power by an amount equal to the power losses .
The ratio of output power to input power gives what is
known as the effi ciency of an electrical machine, defined as

P e/PIll

'1 - 2:.P / (P e

+ 2:.P)

in the generator mode of operation, and as

11 =


'1 - 2:.P/(P m


+ 2:.P)

for the motor mode of operation .

It will have been noticed that the effi ciency is expressed as
a fraction , th at is , on a per-unit basi s. It may as well be
expressed 011 a percentage basis.
The efficiency of an electrical ma chine is le ss than uni ty
on a per-unit ba si s, or less t han '100 % on a percentage basis.
Obviously, as the losses decrease th e effi ciency approach es
unity (or '100%) .
To prevent overheating, the heat dissipat ed in a machin e
mu st be withdrawn a nd disch arged to th e surroundings b y
a cooling system using a gas (mo st Irequent.lv, air) 01' a liquid
as the cooli ng agent.

Electrical Losses

The electr ica l losses in a m achine can eJfect iv el y be redu ced

by making its conductors of a material having a low resistivity , Pt . The best choice is sof t coppe r wire of circula r or
rectangular cro ss-section with a low impurity content.
Accordingly , this kind of l oss is trad i Lionally called th e
copper loss. The second best choice is al um in ium which is
cur ren tly used on a limited , bill, an ever increasing sca le .
Its resistivi ty is, how ever, mu ch higher than tha t of copper.
Bec au se the wind ings carry al terna ting current, we h ave
to reckon with the skin effect. It gives rise to variations in
the inductive impedance and , as a consequen ce, in th e
distribution of current density over the cress-sec tion of
conductors. Th ese variations ar e more noticeable in the
conductors laid ou t in slots than when they are surroun ded
by a nonmagnetic medium , say, ail'.
This loss is well known FR loss , but R must be the effec tive resistance. The mea sured d .c. resi stance is only th e
effecti ve resistance at low frequencies, when the current


CII . 31 En ergy Conversion Losses

dis tribution m ay be assumed to be uniform

Ro =

where 2wl m ean

2wlme an

P t-~=



= leng th of series-connecte d wind ing (or

pha se) con duc tors

lm ean = m ean length of a half-turn
(as b s) CaCb = cr oss-sect ional ar ea of th e
effecti ve conductor
a s, b s = dimensions of a rectangular
strand in the slot height and
width ,
r especti vely
Fig . 31-1)
Ca , Cb = number
of strands in the
slo t height and wid th , respecti vely
a = number of circuit s in the
Pt= P20 [(1 ex (t - 20)] = r esistivity of the conduc tor at
the design operating tem pera ture t
ex = 0.00 4C- 1 = temperature
resistance for copp er (or aluminium)
For a. c. the copper loss has to be compu ted in te rms of
R = k RR o, the resistance of th e winding with allowance for
a nonuniform curre n t distribution over the cross-sec lion of
the con ductors .
The extent of variations in the a .c. di stribution over the
con duct or cr oss-section depends on the magnitude of the
slot leakage field. Becau se the lines of that fi eld ar e at right
angles to the slot axis and are nearl y s tra ight lines in a rec tangular slo t (see Fig. 31-1), the slot le akage flux has just
about the same linkage with any strands l ying at the same
level in t he slo t height (say, str and s 1 and 2). Accor dingly,
th e inducti ve impedanc es of su ch stra n ds are t he sa me, t oo.
In contrast, the strands tak ing up different positions in
the slot height differ in inductive impedan ce as well. As is
seen from Fig. 31-'l a wh ere the slot is shown to con tain onl y
one conductor, the inductive imped anc e (or flux linkage) of
s tran d 1 which is nearer to the air ga p is sm aller than that
of strand 3 lying closer to the slot bottom . This also explains
\\ hv the curren t is dis tri Luted almost uniforml y across thE.'


Pa rt Two. Energy Conve rs ion by Electrical Machines

conductor width and less uniformly along the conduc tor

height. 'When a slot contains only one conductor , th e cur rent.
densi ty is higher in the s tr ands th a t are nearer t o the air gap
(see the current disLribution curve in Fig. 31-1a). In that part
of the cross-section, the cur rent den sity ma y substantially
exceed the a ver age current density in the conduc tor
J o = IIS = IIa~bs
Th e part of the conductor section lying deeper in a slot
carries onl y an in significant fraction of the total con rl uctor

Fig. 31-1 Alternating cur re n t densi ty (J ) distribution over the cros ssecti on of an effect ive con duct or
(a) one-piece effect ive con duc tor, Cn = Cb = 1, lt c = 1; (b) effective
conductor subdivided into st ra n ds transposed in the slot depth , c" =
10, cb = 2, lt c = '1

current. As a result, t he useful cross-section of the conduc tor

decreases, and its resist.ance goes up. This property is accounted for by wha t is known as the Fi eld coeffic ient defin ed

= ius,

As is seen from the foregoing, it is a function of th e heigh t

and number of u nstranded effec tive conductors per slot,
and is independen t of their width.
Consider the most common forms of slot conductors.
1. Unstranded effective conductors. A slot with a single
effective conductor which consists of 'only one s trand occupying the entire slot dep th (ea - 1) is shown in Fig. 31-1a.
The height of the effective conductor is the same as that


Ch. 31 En ergy Conversion Losses

of the strand as. The number of effective conductors in the

slot depth is the same as that of strands U c = m s = 1. The
number of strands in t he slot wi dth, Cb, may be taken such
that the total width of strands in the slot is b, = cbbs.
(What is important is that t he effective conduct or is not
stranded in the slot depth.) With this arrangement the
magnitude of the skin effect is a fun ction of the rel ative
height of a strand
= a s / fJ.
which is defined as the ratio of t he height of a strand as, t o
the skin depth * defined as
fJ. = V2ptbs/CUflob1
where bs = slot width
bl = cbbs = conductor width in the slot
co = 2rr,j = angular f zequency of the current
flo = permeability of the conductor material (copper
or aluminium).
Assuming that the skin effect is onl y observable within
the active length of a half-tun: (the part enclosed in a slot) ,
that is , over the length lfll and is non- existent in the
overhang, that is, over the length lmean - I f' ' we may wr it e
for k R
k R = '1
(l o/Zmeari)( k R a - '1)
wh ere

kR a

= cP

m+ ~ '1)(5) (k~m~ -



7)/16 is the chording (pitch-shortening)

where kiJ = (g~
factor (~ = ych:) for double-layer windings (for singlelayer windings, k~ = '1) , and cp () and '1 ) () ar e th e Emcle
functions (see Fig. 3'1-2). For ~ '1,
cp () = 1

5> 2,



'1) () = 4/3
cp () ==

'p () = 2
* For a conductor carrying currents at a giv en frequency as a result of the electromagnetic waves incident on its surface this is th e
depth below the surface at whi ch th e current density ha s decreased
one neper below that at th e surface.-Transl at or's not e.


Part Two. En ergy Conversion by Electrical Machines

As is seen , the va lue of k R (and the winding loss) in creases

with an increase in the relative conductor height and the
numb er of effective conductors in a slot. As a consequence,
when the effective con duct or in a slot is unstranded (ca = 1),
the loss may be pro hibitively heavy. A way out is to strand
2.0 If

1. 6






Fig . 31-2 Emd e fun cti ons

it (in the slot depth) and to transpose the strands within

. t he slot (case 2 below) or within the overhang (as in case 3
below)* .
2. Stranded effective conductors transposed within the
slot. Ref erring to Fig . 3'1-'1b , each turn of th e effective conductor is made up of two bars joined (by sol deri ng, brazing
or welding) at the ends and completely transposed within
the slot. The construction of the bar is clear from Fig. 3'1 -3.
Owing t o the tr ansposit ion , each stra n d suc cessi vel y takes
up all th e likely positions (or levels) in a bar. As a result,
the strands in the effective bar have the same inductive
impedance , and the current in the effect ive bar is equally
share d among all the strand s


IIcac lJ

The current distribution may be other than uniform only

within a given strand (see the J curve for strands). The nonuniformity is notice able in the str and sections lying nearer
to the air gap where the leak age field is stronger. Even then,

St randing without transp osition would not redu ce th e losses. ,


Ch, 31 Energy Conversion Losses

the distribution in such sect ions is more uniform than it is

in an unstranded effect ive conductor (compare Figs. 31-1a
and b). The current densi ty at the peripher y of a strand
diff ers bu t little from the av erage current density

r, =

Is /asb s

Therefore, the losses in a t ra nsposed str anded effective

conductor are subs ta n tia lly smaller than they are in an

2 34





Fig. 31-3 Bar conductor in which the strands are transposed within
the slot

unstranded cond uctor of the same cross-sectional area. In

such a case, k R can be found from Eqs. (31-3) through (31-6).
The number of strands in the slot depth is
where ue is the number of transposed effective conductors
in the slot depth (as a rule, U e = 2).
3. Stranded effective conductors with strands transposed
within the overhangs. The strands are transposed by twisting
some of the effective conductors. Figure 31-4 shows a doubleturn coil with two strands per effective conductor (ca = 2,
U c = 2). The strands are electrically joined (by soldering,
brazing or welding) at the coil ends. Within the coil, they
are insulated from each other. The inductive impedance of
t he strands (say, 1 or 2) depends on the position they t ake up
I n the depth of the slot where the coil is laid. In an "untwist:<: 5 - 0 16 9


Part Two. Energy Conversion by Electrical Machines

ed" coil (shown at the upper left of Fig . 31-4) an d in a

"twisted" coil (shown at the bot t om ri ght 'of Fig. 31-4) where
the effective conductor is turned 180 0 every turn, the strands
t ake up different combinations of posi tions.

Fig. 314 Arrangement of strands in t he effective conductors of a

coil. The sketch at the bottom right shows a part of a coil overhang
with some of the st rand twist ed (tr ansp osed). The numerals in parentheses refer to the corresponding strands in the twi sted coil after each

In a "twisted" coil, t he difference in inductive impedance

between the strands is less noticeable than in an "untwist ed"
coil, and the current distribution among the strands is more
uniform. For such a coil, k R is given by

,k R = cP (~c)

+ N (~c) /2 + L s (~4/6c~)

(lmean/lrJ (as/a;)2 (31-7}


Ch. 31 En erg y Conver sion Losses

= rela t ive he ight of the effec tive

where Se .= ca S ~V/" as ~il6
mea n
con ductor
rp (Se), cP (Se) = the Emde func t ions for e in
Fig . 31-3
as = height of a bar e strand
a, = height of an insulated strand
L, L s = coeffic ients to be taken from the table
according to type of transposition.
Coefficien ts Land Ls

Coil t y pe

Twist ed eve ry


u~- 1 )
-1 ( 2



3)+ 61



( ue2kfl'

Magnetic Losses

Magn etic losses in the cores, or the core loss of ele ctrical
m achines occur owing to periodic variations in t he magneti c
field with t ime.
Here, too , the core loss can be minimized by subdividing
the core into electrically insulated elementary magneti c
circuits. Th e required effect ive cross-sectional area of th ecore is obt aine d as the sum of the cross-sect ional areas of
the elementar y magn etic circuits whi ch t ake the form of
ferromagn etic lami na tion s in sulated from one ano ther and
m ade in certain thicknesses . Th e material and thickness of
t he lamin ations are ch osen accordin g to the frequency of
cycl ic magn et ization.
As has heen shown is Sec. 21-2, the frequency of cyclic
magnetization for t he st ator is diff eren t from t hat for the
rotor in the general case (WI =1= w z* ), eac h fre quency bein g

* Th e referen ce is to the most typica l a.c . mac hi ne in whic h oneof the win dings is laid in slots on th e sta t or core and t he other in
slots on the rotor core. If the windings are carried by th e same core ,
two magnetic fields will exist within it , each var yin g at a frequency
of its 0'""11, WI and 002'


Part Two. Energy Conversion by Electrical Machines

the same as that of the current in the respective winding.

In induction machines, the relation between the two frequencies depends on the rotational speed, and the lamination
thickness must be chosen to suit the nominal speed of rotation. To obtain a uniform distribution of the magnetic
flux over the cross-section of each lamination and to keep
the core loss to an acceptable level as the frequency is increased, the laminations are made progressively thinner and
from suitably alloyed electrical-sheet steels. The cores to be
cyclically magnetized at about 50 Hz (which is true of, say,
the _s tat ors of synchronous and induction machines) are
assembled with laminations punched from hot-rolled electrical-sheet steel, usually 0.5 mm thick. The more recent
trend has been towards cold-rolled, nonoriented-grain electri-cal-sheet steels which show a reduced specific iron loss. In
large machines, the poles are fabricated from cold-rolled,
grain-oriented steels having still better properties (specific
loss and permeability) when magnetized in the direction of
rolling; such punchings are made also 0.5 rum th ick. For
higher frequencies (400 to '1 000 Hz), use is made of highalloy electrical-sheet steels in thicknesses of 0.35 mm,
,0 .2 mm , O. '1 mm , and 0.05 mm. For machines where the
frequency of cyclic magnetization is several hertz or a fraction of a hertz (such as in the rotors of induction machines),
or is zero (such as in the rotors of synchronous machines
where the magnetic field remains constant in magnitude
and direction) the cores may be assembled from structuralsteel laminations. The thickness for such laminations is
usually chosen from manufacturing considerations (ease of
stamping) and may be '1.0 mm , '1.5 mm , 2.0 mm , 4.0 mm,
,6 .0 mm , and more.
As often as not, especially in mechanically strained rotors,
the cores are made in one piece from steel forgings or steel
(sometimes cast-iron) castings. In more detail, the properties
'Of electrical-sheet and structural steels are discussed in ['13].

(i) Cyclic Magnetization of the Core of an

Electrical Machine
Figure 3'1-5 shows the rotating magnetic field set up by the
currents in the three-phase winding of an electrical machine.
If we compare the field patterns spaced a quarter-cycle apart,
we shall see that the field changes differently in the differ-


Ch. 31 Energy Conversion Losses

[a )


8 1m



0(8) ( e)


1 7

8 1m

J 5






.1 7

2 Ii


(d )





I 7
2 Ii



0(6) 7

2----.0---- Ii
If !i

Fig. 31-6 Magnetic flux density for various forms of cyclic ma gnetization:
(a) pul sational magnetization; (b) rotational ma gnetization; (c) mixed or elliptical magnetiz ation (with the magnetic flux density vect or
represented by the sum of two pulsational vectors) ; (d) mixed or ellipti cal magnetization (wit h the m agnetic flux density vector repr esented by the sum of a rotational and a pulsational vector)


Part Two. En ergy Conversion by Electrical Machines

ent parts of the magnetic circui t . In t he te eth (say, at

point I) , the field lines are always directed radiall y, and the
f ield only changes periodically in m agnitude as in a transformer. Accordingly , we may call it pulsational or transform er magnetization . In Fig. 3f-Ga , the magnetic flux density
vector with a peak value B l m for pulsational m agnetiz ation
is shown at instants 0, '1 , 2, . . . , 8 spaced TI8 ap ar t.
In the yoke of the inner core, the m agneti c fiel d changes
in a different manner. At point, say, II in Fig. 3'1-5, it
r em ains un chang ed in m agnitude and onl y cha nges in
direction. For this field, the magne tic flux den sity vector
rotates at an angular velocity co = 2nf r elative to the yoke,
while retaining its value . Accordingly , we ma y call it
rotation al magnetization*.
In t he case of rotational magnetiz ation , cha nges in t he
magnetic flux density vector are shown in Fig. 3t-6b . Now
the vector B can be depicted as the sum of two pulsa t ionaI
magnetic flux density vectors equal in peak value to the
magnitude of the ro tating B vector, that is , 1B I. To t his
end, it will suffice to displace the ve ct or axes by n /2 and t o
cause one of them to lag behind the other by a qu arter of
a cycle or an angle n/2 (so t hat one of the pulsational vect ors
is a maximum and the other vanishes).
In the outer yoke , the magnetic flux density h as both
a radial and a tangential component , su ch that at points
where at a given instant the tangential component is a m aximum (at say, point IV in Fig . 3t-5a, wh ere B ", = B l m ) ,
the radial component is nonexistent iB, = 0). Conversely,
a t points where at a given instant t he ra di al com ponent
is a ma ximum (at, say, point III in Fig. 3t-5a, where
B r = B z m ) , the tangential com ponent is nonexiste nt
(B ", = 0). A quarter of a cycle later (see Fi g. 3t-5b) , t he
r adial component becom es a maximum at point I V (B; =
= B 2 m ) , and the tangenti al component vanishes (B ", = 0).
The reverse cha nge occurs at poin ~ I I I (B l' = tIm , and
B; = 0).
The relative magnitudes of the radial and t an gentia l components, B l m an d B 2 m , va ry from point t o point roun d t he
outer periphery of the yoke . The peak val ues of both are
maximal (B l m = B 1m ax an d B z m = B zn,ax ) on the inner

* Rotat ion al mag neti za tion occurs in th e yoke of t he inner core

only in the case of a t wo-p ole field (p = 1). Wh en p > 1, ellipti cal
magnetization t akes pl ace in th e inner core (see below).


Ch. 31 Energy Conversion Losses

periphery of the yoke. As the radius of the yoke is increased,

both components decrease in peak value , but B l m is reduced
insignificantly, whereas B 2 m on the outer periphery of the
yoke vanishes (if we assume that the field is confined within
the limits of the yoke). Thus, on the outer periphery of the
yoke, where only the t an gen t ial time-varying component
exists , with a peak value B l m , the magnetization is puIs ational. On the inner periphery of the yoke, where two puIs ational components unequal in peak value exist , displaced
from each other in time and space by an angle n /2, the
magnetization is elliptical. Now the magnetic flux density
vector not only rotates at a frequency whose mean value is
<U = 2nt, but also changes in value from B l m to B 2m. It is
seen from Fig. 31-6c that the tip of the magnetic flux density
describes an ellipse in this form of magnetization.
In the case of elliptical magnetization, the magnetic flux
density vector may be written as the sum of two pulsational
B = qxBlm cosor

+ q yB 2m sincoz


where B l m is the peak value of the vector pulsating along the

x-axis and B 2m is the peak value of the vector pulsating
along the y-axis.
As already noted, the locus of the B vector in the general
case (for B l m:::/= B 2m) is an ellipse. In vector form, the
ellipse is described by Eq. (31-8) where the parameter is
time, t. For rotational magnetization, which is a special case,
with B l m = B 2m, the locus of the B vector is a circle. This
form of magnetization is quite aptly called circular. For
pulsational magnetization, which is another special case where
B 2m = or B l m = 0, the locus of the B vector degenerates
t o a straight line which is aligned with the x - (or y-) axis.
Elliptical magnetization may be treated as a hybrid form
because it can be visualized as the superposition of rotational
and pulsational magnetization. On re-writing Eq. (31-8) as

qx (B l m -

B 2m) cosroz

+ [qxB2m cos <ut

+ q yB 2m sin rot]
we can see that the magnetic flux density vector is obtained
as the sum of a rotating vector whose magnitude is B 2m
and a pulsational vector whose peak value is B l m - B 2m.
Exactly this form is given to the field in Fig. 31-6d:


Part Two. Energy Conversion by Electrical Machines

(ii) Core Loss with PulsationaI and Rotational

The core loss associated with pulsational magnetization
was considered in connection with transformers (see Sec . 2-7).
The core loss associated wi th rotational magnetization differs
from that of pulsational magnetization . A comparison
of hysteresis loss for rotational and pu lsational magnetization is given in Fig. 31-7. For rotational magnetization, the
loss is plotted as a function of
the magnitude of the B vector,
for pulsational magnetizaPM
as a function of the peak
1/ \
va lue of the B vector.
At B < 0.7 T, the core is
slightly saturated (the perme3
V /
ability is nearly constant), and
may treat rotational mag we
netization as the superposition
of two independent pulsationT
magnetizations along two
F ig. 31-7 Hysteresis loss in (1)
rotational magnetization and (2) mutually per pen dicul ar axes .
For the va lue of B given above ,
pulsational magnetization (for
the hysteresis l oss assoc iated
1.91 % Si)
with rotational magnetization
is ab out twice as gre at as that
for pulsational magnetization. As the value of B is increased,
the core is saturated pr ogressi vely mo re, and the pr inciple
of superposition is no longer ap pli cable owing to a substantial nonlinearity of cyclic m agnet izat ion .
As is seen from Fig . 31-7, the nonlinearity of the function
B = f (H) at B > 0. 7 T ma nifests itself in that the ratio of
hysteresis loss in rotationa l magnetizat ion to t hat in pulsational loss gradually decreases so that at B = 1.0 or 1.5 T
it is from 1.45-to-1 to 1.65-to -1. When B is about 1. 7 T,
hyst eresis loss is the same in either case, whereas a further
increase in B leads to a sudden decrease in hysteresis loss
associated with rotational magnetization, so that it is a fraction of that in pulsational magnetization .
Eddy-current loss is solely a function of the magnetic
flux density in the laminat ions and is independent of the
field intensity. If we write the magnetic flux density in the
case of rotational magnetization as the sum of two pulsation-

Ch. 31 Energy Conversion Losses

al components, we shall see t hat the eddy-current loss in this:

case is twice as great as the eddy-current loss in pulsational
magnetization for the same peak value of magnetic Ilux.
density (that is, irrespective of the magnetic flux density) ;
(iii) Magnetic Loss in the Magnetic Circuit Elements

In calculating the core loss in the cyclically magnetizedelectrical-machine components assembled with elecrical-sheet
steel laminations insulated from one another, it is important.
to consider the form of magnetization (pulsational or rotational) , the increase in the loss due to manufacturing factors;
and also various additional losses.
The point of departure in determining the magnetic loss.
in magnetic-circuit elements is the total loss in 1 kg of laminations, assuming pulsational magnetization at 50 Hz and
a magnetic flux density of 1 T. The total loss is measured bywhat is known as the Epstein apparatus and referred to as
the specific loss designated by PLO/50' The values of PLO/5 0'
for various steels are given in [13] .
For other values of frequency and magnetic flux density
(B~ '1.6 T), the specific loss can be found by the equation,
Pm =




if f ranges anywhere between 40 and 60 Hz. If, however, f va-ries over a broader range, the specific loss is found by aIL
equation in which the hysteresis loss and the eddy-current
loss are separated . This is usually accomplished by taking'
advantage of the fact that for a given value of B m , thehysteresis loss varies directly with the frequency and the eddy-current loss with the square of the frequency. This.
equation is
Pm =

e. (lISa) B2



(31-10) -

where e = specific hysteresis loss, W /kg, at B = 1 T and

f = 50 Hz
(j =
specific eddy-current loss, W /kg, at B = 1 T '
and f = 50 Hz .
The specific core loss is measured under carefully controlled conditions. Among other things, it is reqnired that theindividual core laminations should be ideally insulated from.
one another and annealled after cutting or punching, and.


Part Two. Energy Conversion by Electrical Machines

.suhject ed to a sinusoidally varying magnetic flux density.

In commercial ma chines , however, the workmanship is not
'so perfect. In most cases, the laminations are not annealled
aft er cu t ting and punching and this lead s t o an increased
hysteresis loss. The insulation between the laminations is
.oft en damaged by t he heavy pr essure us ed in the assem bly of
cores. Nor is it alw ays possible to avoid electrical contact
be tween the laminations and the frame (or shaft) and also
through burrs on t he core teeth. Because of t his , addit ional
.shor t-circuit ed pa ths are formed for eddy currents. This
increase in core loss owing to manufacturing fac tors is accou nt-ed for by applying sui t able correction factors.
In calculating t he loss in the v ar ious elemen ts of the
magnetic circuit , one has also to r eckon wi th the fact t ha t
the magnetic flux density in the case of pulsational magnet ization varies non sinusoidally (whereas in core-loss
.measurements by t he Epstein apparatus , it is made to var y
.sinusoida lly).
If we expan d th e magnetic flux density into a Fouri er
-ser ies, we shall see that, in addition t o the fundament al
-com ponent, the seri es also contains higher harmonics . The
.losses due to the higher harmonics are added to those asso ci a.t s d with the fundamental component , an d may quite aptly
.be called additional core losses. If they are not calc ulat ed
. .separ at ely , the additional core losses are accounted for by
.applying addi tional-loss coefficients.
I t should be no ted that the additional magnetic losses
-can be caused even in the cores whose windings carry cur.rents at zero frequency (that is , direct current). The funda:m en t al magnetic flux density in such cores (for example, in
t he poles of synchronous machines) varies at zero frequency
.a ud, as a conse quence, t he basic core loss is non-existent. In
.contrast, the addi tional losses associated with the higher
.harm onics of ma gn etic flux density, varying at a high fre .qu ency, may be considerable. This is especially so when
one core is assembled of heavy-gauge laminations or even
nna de one-piece (because there is no cyclic magnetization at
the fundamen t al frequency), and the other core (on the other
.side of the air gap) ha s an appreciable saliency which gi ves
ri se to no tice able pulsations in t he magnetic flux density
-on its surface. For such cores, t he addit iona l magnetic
.Iosses should be calc ula t ed separately and added to the to tal

C h, 31 Energy Conversion Losses

~ iv )


Magnet ic Loss in the Core Yoke

In calcula t ing th is loss, remember that the m anner of

magnetization is ell ip tical at t he boundary with the toothed
I a yer , and pulsational on the peripher y. For t he yoke, the
-over all coefficient allowing for t he increase in loss due t o
in accuracy in manufacture is taken as kad a = 1.3 to 1.6,
.an d the yoke loss is given by
(31 -11)
where ln a = mass of the yoke iron; Pma = spe cific yoke
l oss by Eq . (31-9) or Eq. (3'1-'10) as m easured at B a and
rthe fundamental frequency i: B a = peak value of the
'tangent ial magnetic flux density in the yoke as found for
'the fun damental magnetic flux densi ty b y ca lcula t ion of t he
magnetic circuit for a given type of electrical machine (see
P ar ts 4 and 5) .
( v ) Magnetic Loss in Core Teeth
'The cyclic magnetization of core teeth is pulsational. Therefore, account needs to be taken only of the in crease in losses
owing to manufacturing fa ctors and of the higher time
h arm onics . As com pared with the yoke, the teeth are sma lle r
i n siz e, and the effect of cutting or stamping is noticeable
o ver a larger too th area . So, the coeffi cient t aking care of
t he increase in loss due to manufacturing factors is larger
t han it is for the yoke . In addition t o the fundament al
c om ponent of peak value B z- the magnetic flux densi t y in
a tooth has substantial harmonics . Therefore, the additional
'loss coeffi cie nt for t eeth is k a d z = 1.7 to 1.8 . The t oot h
l oss is given by
(31 -12)
where In z = mass of the too th iron; Pm. Z = specific too th
l oss as given by Eq . (31-9) or (3'1-'10) at B z and i, B z =
pe ak value of ma gnetic flux densi ty in the mean tooth
sect ion for th e fund am ental com ponent , as found by calculat ion of the magnetic circuit for a given machine (see
Parts 4 an d 5); f = frequency of cyc lic m agne tization
for t he fund am ental m agn et ic fl ux density.
As in transformers, t he core loss is m ainly a fun ction of
t he mutual flux. As is seen from E qs. (31 -9), (:1'1 -it) and


Part Two. Energy Conversion by Electrical Machines;

(31-12), the core loss is proportional to the voltage squared'

and remains nearly constant as the winding currents vary.
The losses associated with the cyclic magnetization by'
winding leakage fields are poportional to the square of
current . These losses are lumped with the additional load
loss which is a function of the load current.

Mechanical Losses

These consist of hearing-friction loss, brush-gear loss, win-

dage loss , and cooling loss.
Bearing-friction loss. Its magnitude varies with bearing
design and the lubricant used. In small machines, the
bearing-friction loss is kept to a tolerable level by using
grease-packed ball and roller bearings. In larger machines,
preference is given to sleeve bearings and thin lubricating
oils . With all other conditions being equal, the hearingfriction loss builds up with increasing rpm, rotor mass, and
shaft diameter at the bearings .
Windage loss . Its amount increases with increasingdensity and viscosity of the medium inside the machine.
The loss is a maximum when the rotor rotates in a
liquid medium, and falls to a fraction when the rotor rotates in air. A further decrease by a factor of about ten is .
obtained when hydrogen is used to fill the insides of a machine.
To minimize the windage loss, the outer surface of the
rotor must be made as smooth as practicable . For a given
power rating and a given rpm, the 'win dage loss is larger in
machines with a larger rotor diameter (with a larger ratioof rotor diameter to design length).
Cooling loss. This loss is the power spent to drive fans or
pumps supplying a circulating coolant in the cooling system
of a machine. It is proportional to Q, the flow rate of coolant,
and H, the head developed by the fan or pump, and increases with decreasing efficiency of the fan or pump.
I t can be reduced by making the hydraulic system as perfect as possible and by choosing an appropriate coolant. For'
liquid coolants it is lower than for gases because a liquid
coolant has a higher specific heat capacity, and its flow rate
may be kept at a lower level.
The cooling loss is found as explained in Part. 3.


'Je n era l

:.1.' Il erpcn r . H. 9 JlenlllpU 1l eCnue suuuuu u. 8rrepruJI, Mocxna: toI. I,

1974; ;q . 2, 1963; toI. 3, 1968.
'2 . Hocrerrxo M. Il . 9.MnmpU'IeCnue siauuucu , f[ acme cneuu a.uaui s:
'1 I'o carrepronanar , Mocxn a-Jleumrrpan , 1949 .
.'3. Kostenko M. P ., Pi otro vsky L. M., Electri cal Ma chines. l\Iir
P ublishers, Moscow, 1973.
4 . :BOJIbAeK A. H. 9J1enm p ll lechlle sia uuuc u , 8 HeprIIJI, Jl enrm r pan ,
1974 .
.5. :Banm on A. H. 9.!/,ChmpU'leCnne suuuuicu, 8rrepruJI, JIeHIIIIrpaA,
1974 .
'6 . Cepreen n. C. 9.;.ehmpn
e suuu u n u . rOCi:lIIeprOU3AaT, M ocxaa-Jlemrtrrpan, 1962.
7 . AJIeKCeen A. E. H oncm.punu u sx enmp iocecnux suuuu.n . I'o cane proII3AaT, 1958 .
8 . Bnnorpanon H. B. Il p ouseoiicmeo es enmp uue cnux suuuun., 8rrepruJI,
Momma, 1970 .
9. Beprnaoa A. H. 9J1e/;mpll'leChUe siatuuu a aeuauuoiut oii aemouamunu , Ofioponrna, Mocxsa, 1961.
1 0. Xpymea B. B . 9J1ehmpU 1lecl>u e suutposuuuiuui aemosuunu uecnux
ucmp oiicme, 8UeprIIJI, JIeHIIHPpaA, 1976 .
11. I0cPepon <D.M. 9 J1el>mp u'lecl>ue suuuuu. u aemouamuuecnux ycmp oiicmo.
BhICIIIaJI IIIKOJIa, Mocxna, 1976 .
12. Bpycxnn JJ:. 8., 30pOXODII'I A. E ., Xuocron B . C. 9J1e/;mpu'lec/;ue
xuuuuiui U suucposuiuuut u ; BbICIIIaJI IIIKOJIa, Mocrcaa, 1971.
13 . rpYAIIucRmI n. r. (pen .) 9J1.e/;mpomexllu'lec/;uz'i cn.pa eouuun, T . 1,
8HeprnJI, Mocxsa , 1974 .
14 . Korn, G. A., Korn, T. IlL, Mcthemattcai Handbook fo r Scientists
and Engineers, McGraw-Hill Book Company, New York, 1968.



Part One

15. Ilerpon r . H . Tpallcrjiop .ltamOpbl. OHTI1, Mocxna, 1934.

16 . TIIXOMUpOB n. M. Pncueni m pancqicpsui m o poe, 8HeprrrJI, Mocxaa,
17 . CarrOJImrrROB A. B. K oncmptj upoeauue mpallcrjiop .ltamOpoo. roci:lHeprOII3AaT, Mocxsa-Jlerraarpan, 1959 .
18. AJIeKCeeHKo r. B., AIIIpJITOB A. H ., Bepexreii E . B., <DprrA E . C.
JI cnurnanusi sunuuu m p aucifiop uamapoe U peanmopoe. 8Heprmr,
Mocxaa, 1978.
19. Hararroaas E. A. Hcnumauue mpaHcrjiop.l1amopoo .il1aJloz'i U cpeiineii scoiuuocm u ua tcanpscucenue ao 35 n.B Oh".JlIO'lUmC/!bIlO. 8rreprrrJI,
Mocnaa, 1969.

Bibliograph y


20. cDap6Man C. A . , Byu A. 10 ., Pa iixnn n H . ~ l. P esionni II sioilep-:

n u ea uusi m pancifiop w am op oe , Mocxaa, 8 uep rllH, 1974
21. BaCIOTJInClHIii C. E. B onp ocu m eopuu u. p acuem a m paucrPop.llam opOB. JIeHIIHrpaA, 8 ueprIlH, 1970.
To Part Two

22 . II OJIIIBan OB K. M. T'eop em unecxu e OC/WBbl ssesnipomexnu.n u, q. 1Jlnaenssre 8JIeliTpJIqeCKJIe lJ,eIIII C cocpenor ose nnsnnr IIOCTOHnnbIMII . 8BeprIIH, Mocxaa , 1972.
23. JRYXOBJIIJ;KIlii E . H ., Hernea nmmii H. E. Te op emunecsu e OC/IOB bl..
es enmpomeznunu, q . 2. Jlanefimae 8JIeKTpIIQeCKIIe lJ,eIIII (IIPOAOJImenIIe). Hennaeiimae 8JIeKTpnQeCKJIe IJ;eIIJI. 8 neprIlH, Mocxaa, 1972.
24 . IIoJIlIBanoB K . M. T eopem uuecnue OCItOBbl enenmpomexnunu q . 3.
Teopua 8JIeKTpOMarnJITnOrO IIOJIH . 8HeprJIH, Mocxsa , 1975.
25. Adkins , B. The Genera l Th eor y of Electrical Machines, Chapman
and Ha ll, Lond on , 1959.
26. White, C., Woodson, H . El ectromechan ical Energy Conversion _
J. Wiley and Son ., In c., New. York, 1969.
27 . KOllbIJIO B H. II . 8 Aenmp o.H eXaltU eCnoe np eotipasoeanu e suepeuu :
8nepr IIH, M6cKBa , 1973.
28. HBanOB-CMOJIenCKllii A . B . 8AeKmp0l>ta3ItUmltble nOAR. u np ou eccu:
B sne nm.puu ecnu x siauiu n ax u ux rPuaullecnoe .1100eAUp OBaItUe. 8 neprJIH, wlocKBa, 1969.
29 . Schuisky, W. Bere chnurig electrischer M as chinen, Springer , Wien ..
30. Ceprees II . C., BJInOrpaAoB H. B., I'opmnros <1). H . JIp oenmup oeau u e enenm.p uuecn ua: sui uuuc , 8 neprIlH, Mocxaa, 1969.
31. II ocTnIIKoB H. M . tt poenm u p oeau u e 3AenmpU 'leCI>UX .1LaUl!t It. re cTeXJIaAaT YCCP, KlIeB, 1960 .
32. Liw sch it z-Garik , M. W indin g Alterna tin g-C urrent Ma ch in es.
33. KYQep a H., f aIIJI H . Otiscomxu. sne nm puwecn ua: uaiuun , Il epen .
C nemcx . HaA-Bo AKaAeMJIII HaYK, Ilpara , 1963 .
34. 3 mIlIH B. H ., KaIIJIaH M. H ., Ilaaeii A . M. II np , Otiu omm; 3M nm.puuecnu a: suuu u n, 8 nepn JH, Jleanarpan, 1970.
35 . ,I!,aHIweBIIQ H . E ., KarnapcKIlll 8 . f . otiaeo-ucue n ome p u B 3/!8Kmpu uecnu x suuu unax, I'ocaaepronanar, Mocxaa-Jl ea na r pan, 1963.
36 . ,I!,aBIweBIIq H . E ., Kymm 10. A. T eopusi U p acueni oe.unrf!ep/t blx .o6.1Lomon cun xpo n tcux sia uuuc , HaA-Bo AH CCCP, Mocxsa, 1962.
37 . ,I!,anJIJIeBIIq H. E. , )J;oM6pOBCKllll B . B ., Hasoscxnii E . H . Il aposiem.p n. 9Aenmp U'leCIWX suuuun n epesi euu oeo m ona. Hayxa, Mocxsa ..
3 8. Ta JIaJIOB H . H . Il apnsi em p a U x ap anmepucmuxii R.BItOnOA IOClt N:11:
cutcx p ou n u suuuun . 8HeprIIH, Mocxaa , 1978.



Many subjects are not included in the Index because they arelisted separately in the Contents. Therefore, the reader will be welladvis ed first to consult the various sections under the indi vidua l;
chapter headings th ere.
In the Index, each topic has its own section, but subjects common to several topic s may be found separately listed.
A ir gap fac tor, 270
Asyn chronous machine, 23 3
Au to t ransfo rm er , 125, 13 3
Ax ial gap l en gth , 266
Breathing mmf wave , 274

Field windings, 255

l eakage, 45
mutual, 45
Flux linkage, 32 3
coil, 323
coil group, 328
h a rmonic, 33 5
phase , 3 30
winding , 333

Carter coeffic ien t, 270

Coil pitch , 238
Coolin g systems for t ransfo rme r s, 186
Condu ctors ,
Generator, def., 13
stran ded , 381,
t r an spo sed , 384
unstran de d, 382
Imp edance,
Co re
mutual in transformer, 117
cycli c m agn eti zation of, 38 8
short-circuit, 119
sali ent-pol e, 202
Inductance ,
Core l eng t h, 265
l eakage for complete winding , 348;
Cyc li c magnetizat ion , 388
mutual between phases, 343
mutual between stator and phase.,
Distribution factor, 280
mutual in transformer, 51
Effec t iv e core l ength , 265
polyphase winding, 341
El ectri cal machines,
Induction machine, 234
asynchron ous , 233
bas ic arran gement, 18
Load unbalance , 145
ba si c definitions, 13
basic designs, 207
bea ring-friction, 396
cl assific a ti on , 15, 192
cooling, 39 6
disc-typ e, 19 3
copper in transformer, 53
ge a re d , 195
co re, 392
genera l theory, 192
in t ransformer, 53
he teropo la r , 207
ct ri ca l, 37 9-80
h omopolar, 207
magnetic, 379, 387
inducti on, 234
m echanical, 379 , 396
inducto r , 218
n o-load in transformer, 53
lin ear fl at, 19 3
power, 379
linear t ubu la r , 194
win dage, 39 6
r eluctance, 2 14
synchronous , 2 33
Electromagneti c forces, 367
Ma gn etic fi eld,
El ectromagnetic torque, 351, 364
calculation of, 257
from concentrated field winding, 31(;;
rotational , 197
from di stributed fi eld winding, 319'
transformer, 197
leakage, 34 1
Energy conversion,
mutual, 342
by electrical machines , 18
of phase, 267
by rotating m agne ti c field, 372
of pol y ph ase winding, 288
by transformer, 15, 75
periodic, 201
conditions for, 201, 227
from rotating field winding, 316
efficiency, 379 .
spa t ia l pattern of, 262
losses, 37 9
u seful, 341



.Magn et izat ion ,

cycl ic, 388
<p ulsati on ai, 392
rotat ional , 392
Ma gn e ti zin g current;- "!f
Measu rem en t , t ra nsform er quan t it ies,

Ml\IF , 269
phase , 280
Ml\IF eq uation for transrormer . 58
Mot or , def. , 13
P er-u ni t nota ti on, 69

Ph ase belt , 205

.P h ase sequence
negative, 91
positi ve, 91
zero, 99

.P hasor diagra m . 65
Pi t ch fac tor , 276
P ole pairs, 202
,P ol e pi tc h, 23 8
P ower,
el ectri c, 372
electr omagne ti c, 37, 2
magn eti c, 372
'R ot ary convert er, d,ef . , 14
Ro tor, der. , 18
:Self-in du ctance,
comp lete win ding, 3/,5
phase , 31,2
:Slot s per pole, 205
St a t or, def. , 18
.Sur face cur rent , 358
.Synchronous m ac hi ne, 233
' T emperat ur e l imits, t ransform er , 186
'Tempera t ur e rise, t ran sforme r , 18 5
' Test ,
op en-circ uit , 99
shor t-circuit , 102
Tr ansform ati on, three-phase, 79
T ransfor ma t ion ratio , 46
' Tr ansformer, der., 14
arc -wel ding, 180
at no load , 43
.au t o-, 125 , 133
ba sic a rra ngemen t , 15 , 27
bu t t- join t , :l4
<const ru ct ion , 31
cooli ng syst em s of , 186
copper l oss, 53
core an d coil unit , 31
core-an d-shell t ype, 34
...-core type, 33
current, 183
fiv e-leg core-t y pe , 34
imbricated-join t, 34
instrument , 182
phas e-angle error, 183
ra t io err or , 18 3
insula tion-tes ti ng , 181
interleaved-joint, 31,
magnetizat ion curve, 47
ma gnetizin g curr ent , 58
mi tred-joint, 36
.mu lt twin dtn g , 125

no-l oa d cur rent, 49, 117

no-l oad losses, 52
n o-loa d' tes t , 99
on load , 56, 107
open-circuit curre nt , 117
open-circ uit test, 99
ov e rvol tages in, 171
.< .. pal:al!~le d, 138
pa rameter cal culation , 117
peak ing, 182
ph ase displ acement reference nura. bel's, . 82
phasor di agram, 65
shell -t ype, 33
sh ort -circui t impe dan ce, 119 <
short-ctrcur t t est , 102
size re lati ons, 121
spe cial -p urpose, 177
strip-wound, 38
str uctur a l part s, 38
t ap -c h a ng in g , 112
off l oa d, 113
on load, ;I,J.4. _
th ree-ph ase, 8 3, 89 , ]1,5
t ra nsferring t he secondary quantit ies , 62
transients in , 164-
turn s ratio, 46
va ri a ble-volt age, 179
voltage, 182
Transform er core l oss, 53
Tr ansformer emf equa tion , 46
Tra nsformer eq uiva lent circui t , 68
Tra nsformer fittin gs, 4 1
Transformer fra me, 37
Tr ansformer I eads, 1, 1
Tra nsform er load unbalance , 145
Transformer mmf equation , 58
Tran sformer m utu al impedance, 117
Tr an sform er m utual in duct ance, 51
Transformer tan k accessori es, 4 1
Transf ormer t ermin al bus hin gs, 4 1
Transform er v olt age equa t ions, 1, 5, 60
Transformer voltage reg ulat ion , 107
Transform er Wind ing connections, 79
Transformer Win di ng in sulation , 39
Transfo rme r Win dings , 32
Transformer yo ke clam pin g, 36
Triplen harm onics, 91
Win d in gs ,
a .c . ma cbines, 236
ch or ded , 238
conc entrated, 205
cyli nd rical , 202
distri bu t ed , 205
dou ble-l a yer , 20 3
d rum . . 202

fi el d , 25 5
fr acti onal- slot, 250
full-pi t ched, 2 38
l ap , 238, 240
polyphase, 205
select ion of , 246
sbor t-pitched , 238
single-layer , 203
t wo-layer , 236
wave, 238 , 244

Printed in the Union of Soviet Socialist Republics




Professor Alexei V. IVANOV-SMO
~' ', .1." ~~. :'~'.,... , ~ LE NS KY, D. Sc. (Tech.), is a leadin~
"'ilo- ~
. Soviet authority in his field. Currently,
he is with the Moscow Power I nstl.' tute. He has written (individually anc
-a s a coauthor) six books on electrici
ty, including the present one.

-------: ..,.,...-::-+-- - - - -::

~ . '-