# Chapter 1

Electromechanical Energy
Conversion
These notes were used to teach a course in Electromechanical Energy Conversion
in conjunction with the course textbook: Electric Machinery (6th Ed.) by A.
E. Fitzgerald, C. Kingsley, and S. D. Umans, McGraw-Hill, NY, 2003. (ISBN
0-07-366009-4). In some Chapters of these notes there are references to …gures
in that textbook.
1.1 Conversion Process
« Electric or Magnetic Fields in device couple between electro and mechan-
ical domains.
« Various conversion devices operate on similar physical principles
« Structure of a speci…c device depends on its function.
1.2 Device Categories
« Transducers
– for measurement and control
– linear operating conditions for input-output (usually)
– relatively small signals
– e.g.: microphones; pickups; sensors; loudspeakers
« Force-producing Devices
– for actuation and holding- limited range of motion
1
2 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
– e.g.: solenoids; relays; electromagnets
« Continuous Energy-Conversion Equipment
– for energy transfer in large quantities (continuous)
– e.g.: motors; generators
1.3 Purposes of Analysis
« understand process
« develop design/optimization methods for devices with speci…c require-
ments
« develop models for performance evaluation of system components
1.4 Scope of Present Treatment
« Magnetic …elds only are considered for conversion medium.
« Transducers and force-producing devices are considered …rst.
« Continuous energy-conversion devices are considered later.
« For electric-…eld systems analytical methods are similar.
1.5 Force in Electric and Magnetic Fields (Lorentz)
In simultaneous electric and magnetic …elds, Lorentz force on a charged
particle is
1 = ¡(1 + · 1) (1.1)
where
1 = force (newtons)
¡ = charge (coulombs)
1 = electric …eld strength (volts per meter)
1 = magnetic …eld strength (teslas)
· = velocity of particle relative to magnetic …eld (meters per second).
Note that superposition of electric and magnetic forces occurs.
In a pure electric …eld 1 = 0, so by equation 1.1 the force is
1 = ¡1 (1.2)
which acts in the direction of the electric …eld irrespective of any particle motion.
1.6. FORCE BY CHARGE AND CURRENT DENSITIES 3
In a pure magnetic …eld E= 0, so by equation 1.1 the force is
1 = ¡(· 1) (1.3)
which acts in the direction determined by the vector cross product · 1.
Recall that the cross product · 1 de…nes a vector whose direction is
perpendicular to the plane of vectors · and 1 and whose magnitude is
[· 1[ = [·[ [1[ sin0, (1.4)
where 0
.
= angle between vectors · and 1. The sense of direction for · 1
follows from a right-hand screw oriented along the perpendicular. Rotate · into
1 to turn the screw, whereby the vector ·1 points in the direction of advance
of the screw. Alternatively use the equivalent right-hand rule. I.e., point the
thumb and index …nger in the directions of · and 1, respectively. Then · 1
emanates outward from the palm.
1.6 Force by Charge and Current Densities
The following results apply on a "microscopic scale", meaning a volume \
is very small compared to everyday dimensions but large compared to atomic
dimensions. So "\ ÷÷ 0" in a limit expression must be interpreted within
these constraints rather than in the true mathematical sense.
Theorem 1
1
\
= j(1 + · 1) (1.5)
where
1
u
= force density
.
= force per unit volume (newtons per cubic meter)
j
.
= charge density (newtons per cubic meter).
Proof. Consider a small element of volume
\ = ¹[|[ , (1.6)
of cross sectional area ¹ and length [|[ as shown in Fig. 1, which contains
charge
¡ = j\. (1.7)
Assume that this elemental volume is moving with velocity ·, so
1 = ¡(1 + · 1) = j\ (1 + · 1) (1.8)
by substitution of equation 1.7 into equation 1.1. This gives
1
\
.
= lim
\ !0
1
\
= j(1 + · 1). (1.9)
4 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
Theorem 2
1
\
= J 1 (1.10)
is the magnetic force density where
J = current density (amperes per square meter).
Proof. Specialize Theorem 1 for 1 = 0 :
1
\
= j(· 1) = (j·) 1. (1.11)
It remains only to show that J = j· where
J
.
=
i
¹
. (1.12)
Now for the volume element \ the charge density is
j
.
=
¡
\
(1.13)
so
¡ = j\ = j¹[|[ (1.14)
where the last expression follows by equation 1.6; and the velocity is
· =
|
t
(1.15)
if \ moves as part of a current in a continuous charge distribution j. (In this
discussion both i and | are vectors normal to the cross section ¹.) This volume
transports charge ¡ through a boundary surface with area ¹ in time t. This
corresponds to a current i with magnitude
[i[
.
=
¡
t
=
j¹[|[
t
, (1.16)
where the last expression follows by equation 1.14; so
[J[
.
=
[i[
¹
=
j¹[|[
¹t
=
j [|[
t
= j [·[ ==J = j·. (1.17)
where the last step before the arrow follows from equation 1.15.
Remark 3 Equation 1.10 for magnetic force density corresponds to equation
1.3.
The results in Theorems 1 and 2 are useful for analysis in cases where a
large number of charged particles are in motion. The net force on a body can
be determined by integrating 1
\
over its volume.
1.6. FORCE BY CHARGE AND CURRENT DENSITIES 5
Example 4 (Example 3.1- Fitzgerald et al)
See Fig. 2 (RE: Fitzgerald (p. 114) Fig. 3.2 )
A single-turn coil (rectangular loop on non-magnetic rotor) resides in a uni-
form magnetic …eld 1
0
= 0.02 T.
Coil sides are length | = 0.3 m. located at radius 1 = 0.05 m. from rotor
axis.
Coil plane is inclined at angle c with respect to plane surface normal to …eld.
Find torque on coil when coil current 1 = 10 A.
Solution 5 Apply equation 1.10 to determine the force on each coil side. For
one side the force 1 is the integral of the magnetic force density vector 1
\
over the volume of the wire. 1
\
is constant over a side because J has constant
magnitude and direction over the length of the wire on that side and 1
0
is
uniform. Hence the integral reduces to
1 = \ (1
\
) == \ (J 1
0
) = |¹(J 1
0
) = (J¹1
0
) |
where \ = |¹ is the volume of the wire with ¹ as its cross-sectional area. This
gives
1 = (1 1
0
) |
because 1 = J¹ (where 1 inherits the vector property of J with respect to direc-
tion). By evaluation of the above equation for wire 1
1
1
= ([1[ [1
0
[ |) ´ r
where ´ r is a unit vector coincident with the r÷axis, because the current 1 directed
into the paper and the vertical …eld 1
0
are perpendicular. Likewise for wire 2
1
2
= ÷([1[ [1
0
[ |) ´ r
because the current 1 is directed out of the paper. The moment or torque around
the rotor axis is
t = t
1
+ t
2
where
t
1
= 1 1
1
= ÷[[1[ ([1[ [1
0
[ |) sinc] ´ .
and
t
2
= (÷1) 1
2
= ÷[[1[ ([1[ [1
0
[ |) sinc] ´ .
with ´ . a unit vector perpendicular to and out of the paper. I.e., the forces 1
1
and 1
2
are a couple which produces the net torque
t = ÷[2 [1[ ([1[ [1
0
[ |) sinc] ´ .
which evaluates to
t = ÷[2 (0.05) (10) (0.02) (0.3) sinc] ´ . = [÷0.006 sinc] ´ . N m .
Note that the translational forces on the ends which close the loop cancel and
also produce no torque around the rotor axix because both are parallel to it.
6 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.7 Magnetic Force in Electromechanical Devices
All forces due to magnetic …elds in electromechanical energy conversion devices
derive from the Lorenz force given by equation 1.10. Direct computation of such
forces by means of this equation requires the integration of the force density 1
u
over the volume of some component in a conversion device. Here integration
can be interpreted as a sum of in…nitesimal constituent forces; i.e., superposition
applies. This integration requires a detailed knowledge of the magnetic …eld 1
and the current density J as functions of position. The direct integration is a
straightforward problem only if 1 and J are simple functions, as in Example
3.1 where both are uniform.
In general the functions for the …eld 1 and current density J in a device
component are not simple. This situation may be due in part to the presence of
necessary magnetic material which can be analyzed in terms of equivalent surface
magnetization curents due to the net e¤ect of magnetic dipole alignment. In
such cases the forces do not act only on elements carrying free current. This
often occurs by design to meet performance objectives.
In such general cases, the 1 …eld computation (e.g., due to the stator currents
in a motor) invokes a form such as the Biot-Savart equation:
1 =
_
j
0

_
_
\
J (r
0
, j
0
, .
0
) ´a
1
1
2
d\
0
. (1.18)
(Note that this equation follows from the di¤erential form of Ampere’s law,
assuming the displacement current is negligible.) Likewise computation of J
(e.g., for the rotor currents in a motor) becomes complex. Finally there is the
required integration of the force density 1
u
over the component volume. Clearly
this approach is feasible only by means of numerical algorithms.
The energy method is an alternative approach which can be used to deter-
mine the net external e¤ect of the detailed internal Lorenz force distribution. It
applies under the reasonable assumption that device components are rigid (non-
deformable) bodies and enables some simplifying assumptions used for magnetic
circuits. In many situations results are su¢ciently accurate and application is
far simpler than the direct method. The energy method is presented in the next
sectiion.
1.8 Energy Method
The energy method derives from conservation of energy in a system where
some energy is stored in a magnetic …eld. I.e., the magnetic …eld is part of an
energy conversion device and couples the electrical and mechanical portions of
the system. This is shown conceptually in Fig. 3 where a lossless magnetic
energy storage medium appears, to which a pair of electrical terminals and a
pair of mechanical terminals are attached. This can be generalized to include
multiple electrical and mechanical terminal pairs. Note that any electrical and
1.9. MAGNETIC ENERGY BY STATE VARIABLES 7
mechanical loss mechanisms must be extracted analytically from a real mag-
netic device to yield the lossless magnetic energy storage model. Of course loss
mechanisms can simply be ignored if such does not introduce signi…cant errors.
The energy conservation principle for the lossless magnetic storage medium
is
d\
tltc
= d\
ntc|
+ d\
}lJ
(1.19)
where the following represent di¤erential changes:
d\
tltc
.
= electrical energy input = id` = (ci) dt
d\
ntc|
.
= mechanical energy output = ()
}lJ
) dr
d\
}lJ
.
= magnetic stored energy.
The conservation principle also applies to the entire system. In this case it is
necessary only to add the term
d\
loss
.
= loss energy (converted to heat)
to the right-hand side of equation 1.19 if all system energy storage is in magnetic
…elds.The term d\
loss
, which accounts for all system loss mechanisms, includes
those which were necessarily extracted (mathematically) to yield the (ideal)
lossless magnetic energy storage medium. Hence it includes:
1. ohmic electrical losses due to winding currents and magnetic-core eddy
currents;
2. core magnetization losses due to hysteresis;
3. mechanical losses due to friction and windage.
These losses (hence the term d\
loss
) must be represented by electrical
and/or mechanical elements external to the medium. Items 1 and 2 are repre-
sented by resistances connected to the electrical terminals. Item 3 is represented
by dampers connected to the mechanical terminals. In fact even mass (which is
lossless) associated with a moving section of a magnetic core must be represented
by an external mass element connected to the mechanical terminals.
1.9 Magnetic Energy by State Variables
The magnetic stored energy in the lossless medium of Fig. 3 follows by appli-
cation of
d\
}lJ
= d\
tltc
÷d\
ntc|
= id` ÷()
}lJ
) dr, (1.20)
which is a simple algebraic rearrangement of equation 1.19. Here we consider
energy-conversion devices which consist of stationary and moving components so
the magnetic circuit includes moving or varying air gaps. Here signi…cant energy
is stored in the air-gap magnetic …eld which serves a "reservoir" between the
8 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
electrical and mechanical portions of the system. This is in contrast to systems
with …xed geometry (e.g., inductors and transformers), where energy stored in
gap, leakage, and core …elds does not enter directly into the transformation
process.
The magnetic stored energy is
\
}lJ
(`
0
, r
0
) =
_
1
d\
}lJ
, (1.21)
where 1 represents a path of integration from the initial point r = 0, ` = 0
to the …nal point r
0
, `
0
, as shown in Fig. 4. (A path 1 can be de…ned by a
function relation between ` and r.) Equation 1.20 is applied over the path 1
noting that i and )
}lJ
are each functions of ` and r :
i = i(`, r) (1.22)
and
)
}lJ
= )
}lJ
(`, r). (1.23)
The dependency on r enters due to moving device component(s). Magnetic
energy storage is conservative (lossless) here so \
}lJ
(`
0
, r
0
) will be the same
irrespective of the chosen path 1. In other words increments of electrical and
mechanical energy have the same e¤ect on the balance \
}lJ
(`, r) regardless of
the order in which they are added or subtracted by varying ` and r. Hence ` and
r are state variables, meaning that no knowledge of system history is necessary
to determine the stored energy \
}lJ
(`
0
, r
0
) under the present conditions.
The evaluation of equation 1.21 is greatly simpli…ed by choosing 1 as path
2 in Fig. 4.
) \
}lJ
(`
0
, r
0
) =
_
12a
d\
}lJ
+
_
1
2b
d\
}lJ
(1.24)
Here the variations in ` and r are essentially independent over 1
2o
and 1
2b
.
Consider path 1
2o
which starts with ` = 0, maintains d` = 0, hence maintains
` = 0; this implies )
}lJ
= 0 also, because there is no magnetic force without a
magnetic …eld. So
_
12a
d\
}lJ
=
_
12a
[id` ÷()
}lJ
) dr] = 0, (1.25)
by application of equation 1.20, which reduces equation 1.24 to
\
}lJ
(`
0
, r
0
) =
_
1
2b
d\
}lJ
=
_
X0
0
i(`, r
0
)d`. (1.26)
Note that \
}lJ
(`
0
, r
0
) retains dependency on r
0
because the current i(`, r
0
) is
dependent on the …xed r
0
as well as the varying ` during integration over the
vertical path 1
2b
.
1.9. MAGNETIC ENERGY BY STATE VARIABLES 9
Magnetic stored energy is also obtained by integration of the …eld energy
density over the volume \ of the …eld:
\
}lJ
=
_
\
_
_
1
0
H d1
0
_
d\. (1.27)
For soft magnetic material with constant permeability j (i.e., 1 = jH), this
gives
\
}lJ
=
_
\
_
1
2
2j
_
d\. (1.28)
These results follow from …eld theory.
10 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
Example 6 Electromagnetic Relay
See Fig. 5 (RE: Fitzgerald- p. 119, Fig. 3.4 )
The magnetic core 1 ÷ H characteristic is assumed linear and lossless for
simplicity.
The resistance of the excitation coil is extracted to a lumped external resis-
tance 1.
Mass of moving component of armature is extracted to external (lossless)
mass element connected to mechanical terminals.
Any friction loss due to armature pivot is extracted to external damper
element connected to mechanical terminals.
Electrical terminal variables are `, c, and i.
Mechanical terminal variables are displacement r and force )
}lJ
produced
by the magnetic …eld.
)Excitation coil, magnetic core, and armature comprise a lossless magnetic
energy storage medium, in conformance with the requirements of Fig. 3.
By previous analysis of magnetic circuits, where r plays the role of gap
length, we obtain the linear relation:
` = 1(r) i.
I.e., 1(r) is the inductance which depends only on geometry, permeability of
magnetic material, and armature position r.
Solution 7 Now determine the stored magnetic energy as a function of ` and
r, by applying equation 1.26:
\
}lJ
(`
0
, r
0
) =
_
X0
0
i(`, r
0
)d`.
First solve the linear relation above for i :
) i(`, r
0
) =
`
1(r
0
)
;
then substitute this into the equation above for \
}lJ
(`
0
, r
0
), which gives:
\
}lJ
(`
0
, r
0
) =
_
X0
0
`
1(r
0
)
d` =
_
1
1(r
0
)
_ _
X0
0
`d` =
`
2
0
21(r
0
)
.
1.9. MAGNETIC ENERGY BY STATE VARIABLES 11
Example 8 (Example 3.2- Fitzgerald et al)
See Fig. 6 (RE: Fitzgerald- p. 122, Fig. 3.6 )
Electromagnetic Relay with movable plunger is shown.
Core and plunger materials: permeability is assumed in…nite (j ÷÷·).
/
.
= plunger height; q
.
= air-gap length; assume / ¸q.
Dimensions: d = 0.15 mm; | = 0.1 mm; q = 2.0 mm.
MMF Data: i = 10 A.; · = 1000 turns.
==Calculate magnetic stored energy \
}lJ
as a function of plunger position
(0 < r < d) .
Solution 9 Use
\
}lJ
(`, r) =
`
2
21(r)
derived in previous example, but convert this to function of i (given here) by
substitution of ` = (1(r)) i. This gives
\
}lJ
(i, r) =
(1(r)) i
2
2
where
1(r) =
j
0
·
2
¹
¸o¡
2q
,
by previous analysis of magnetic circuits, with the gap cross-sectional area ¹
¸o¡
a function of r :
¹
¸o¡
= |d
_
1 ÷
r
d
_
from geometry of structure. Substitution of this last equation into the previous
equation gives
1(r) =
j
0
·
2
|d
_
1 ÷
r
J
_
2q
and by substitution of this equation for 1(r) into the equation for \
}lJ
(i, r) :
\
}lJ
(i, r) =
_
_
j
0
·
2
|d
_
1 ÷
r
J
__
2 (2q)
_
i
2
.
Finally evaluation for given parameter values yields
\
}lJ
=
_
_

_
10
7
¸_
(1000)
2
(0.1) (0.15)
2 (2 (0.02))
_
(10)
2
_
1 ÷
r
d
_
.
==\
}lJ
= 236
_
1 ÷
r
d
_
J.
12 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.10 Magnetic Force from Energy
\
}lJ
(`, r) is already established as a function of the two (state) variables ` and
r. The di¤erential of any function of two variables can be written in terms of
its partial derivatives with respect to those variables. Applying this concept to
\
}lJ
(`, r) gives
d\
}lJ
(`, r) =
__
0\
}lJ
(`, r)
0`
_
r=constant
_
d`+
__
0\
}lJ
(`, r)
0r
_
X=constant
_
dr.
(1.29)
Both this equation and equation 1.20 must yield the same d\
}lJ
(`, r) in terms
of the di¤erentials d` and dr. Note that this requirement holds for arbitrary
values of d` and dr because ` and r are independent variables. This imposes
the requirement that the coe¢cients of d` must be the same in both equations
and likewise the coe¢cients of dr must be the same in both equations. Hence
we conclude
i =
_
0\
}lJ
(`, r)
0`
_
r=constant
(1.30)
and
)
}lJ
= ÷
_
0\
}lJ
(`, r)
0r
_
X=constant
. (1.31)
Equation 1.30 provides the means to deduce the current i from the energy func-
tion \
}lJ
(`, r). Equation 1.31 provides the means to deduce mechanical force
)
}lJ
from the energy function \
}lJ
(`, r) and can be converted to a function
of i and r by insertion of the function relating ` to i. Evaluation of a partial
derivative with respect to a variable requires that all remaining (in this case
one) variables be held constant. However this is a purely mathematical require-
ment which imposes no requirement at all that such remaining variables be held
constant during actual operation of a device.
Example 10 Any Linear Magnetic System
By de…nition of a linear system
` = [1(r)] i
and in this case the energy is
\
}lJ
(`, r) =
`
2
21(r)
by the results of Example 6. Now applying equation 1.31 to this gives:
)
}lJ
= ÷
0
0r
_
`
2
21(r)
_
X=constant
=
`
2
2 [1(r)]
2
d [1(r)]
dr
;
and substitution of ` = [1(r)] i into this equation gives
)
}lJ
=
i
2
2
d [1(r)]
dr
,
which is a function of i.
1.10. MAGNETIC FORCE FROM ENERGY 13
Example 11 (Example 3.3- Fitzgerald et al)
See Fig. 7 (RE: Fitzgerald- p. 124, Fig. 3.7 )
Solenoid Inductance== Table given as a function of plunger position:
x (cm) 0 0.2 0.4 0.4 0.8 1.0 1.2 1.4 1.6 1.8 2.0
L (mH) 2.8 2.26 1.78 1.52 1.34 1.26 1.20 1.16 1.13 1.11 1.10
Note: r = 0 is position of full retraction.
Plot solenoid force at current 0.75 A. as function of position for range 0.2 _
r _ 1.8 cm.
Solution 12 Apply
)
}lJ
=
i
2
2
d [1(r)]
dr
,
derived in Example 10, where i = 0.75 A. is given here. Hence it is necessary
only to evaluate
J[J(r)]
Jr
from the given table by numerical approximation.
A fourth-order polynomial …t to the data for 1(r) can be obtain by using
the Matlab function poly…t. I.e., we assume
1(r) = a
1
r
4
+ a
2
r
3
+ a
3
r
2
+ a
4
r
1
+ a
5
and poly…t generates the coe¢cients a
|
, / = 1, 2, ..., 5, after the table data is
entered. See textbook pp. 126-27 for Matlab code.
From this
d [1(r)]
dr
= 4a
1
r
3
+ 3a
2
r
2
+ 2a
3
r + a
4
can be substituted into the …rst equation to obtain )
}lJ
. See Fig. 7 for plots:
1(r);)
}lJ
.
14 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.11 Torque from Energy
Consider a system with rotational rather than linear mechanical motion. The
derivation of magnetic …eld energy in this rotational system is analogous to
that in Section 1.9 for systems with linear displacement. Here torque t
}lJ
and
angular displacement 0 replace force )
}lJ
and linear displacement r, respectively,
so the energy balance is
d\
}lJ
= d\
tltc
÷d\
ntc|
= id` ÷(t
}lJ
) d0 (1.32)
in place of equation 1.20. Likewise the mechanical terminal variables in Fig. 3
become t
}lJ
and 0.
We continue the analogy to Section 1.9 to obtain
\
}lJ
(`
0
, 0
0
) =
_
X0
0
i(`, 0
0
)d`. (1.33)
in place of equation 1.26. Likewise
i =
_
0\
}lJ
(`, 0)
0`
_
0=constant
(1.34)
and
t
}lJ
= ÷
_
0\
}lJ
(`, 0)
00
_
X=constant
(1.35)
replace equations 1.30 and 1.31, respectively.
Example 13 Any Linear Magnetic System with Rotational Motion
(This example is analogous to Example 10)
By de…nition of a linear system
` = [1(0)] i
and in this case the energy is
\
}lJ
(`, 0) =
`
2
21(0)
analogous to the results of Example 6. Now applying equation 1.35 to this gives:
t
}lJ
= ÷
0
00
_
`
2
21(0)
_
X=constant
=
`
2
2 [1(0)]
2
d [1(0)]
d0
;
and substitution of ` = [1(0)] i into this equation gives
t
}lJ
=
i
2
2
d [1(0)]
d0
,
which is a function of i.
1.11. TORQUE FROM ENERGY 15
Example 14 (Example 3.4- Fitzgerald et al)
See Fig. 8 (RE: Fitzgerald- p. 128, Fig. 3.9 )
Rotational system is shown.
Magnetic circuit consists of a single-coil stator and an oval rotor.
Coil inductance 1(0) varies with rotor angular position 0:
1(0) = 1
0
+ 1
2
cos (20) ,
where 0 is measured between magnetic axis of stator coil and major axis of
rotor.
Note second-harmonic variation of 1(0) , which is consistent with fact that
1(0) is the same for 0 ÷÷0 + 180

(rotor rotation through 180

)
Given parameters: 1
0
= 10.6 mH; 1
2
= 2.7 mH.
Find torque as function of 0 for current of 2 A.
Solution 15 By the results of Example 13
t
}lJ
=
i
2
2
d [1(0)]
d0
.
Here
d [1(0)]
d0
=
d [1
0
+ 1
2
cos (20)]
d0
= ÷21
2
sin(20)
so
t
}lJ
=
i
2
2
(÷21
2
sin(20)) .
Evaluation for given parameters:
t
}lJ
(0) =
(2)
2
2
_
÷2
_
2.7
_
10
3
¸_
sin(20)
_
==t
}lJ
(0) = ÷1.08
_
10
3
¸
sin(20) N m.
Note that torque acts in direction to pull rotor axis into alignment with coil axis.
This alignment maximizes coil inductance.
16 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.11. TORQUE FROM ENERGY 17
18 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.11. TORQUE FROM ENERGY 19
20 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.12 Coenergy
The magnetic coenergy (complement of the magnetic energy) is related to the
magnetic energy on the path 1
2
in Fig. 4 that led from equation 1.24 to 1.26.
I.e, we assume that the position r
0
is attained by following path 1
2o
before any
magnetic energy is stored in the …eld. Next magnetic energy \
}lJ
is acquired by
the …eld as the ‡ux increases from 0 to `
0
along path 1
2b
while position is held
constant at r
0
. Fig. 9 is a detailed illustration of the latter process, where ` vs.
i represents a general function or characteristic seen at the electrical terminal-
pair of the system. Each characteristic ` vs. i shown in Fig. 9 corresponds to
a …xed value of the position parameter r.
The magnetic coenergy is
\
0
}lJ
(i
0
, r
0
)
.
=
_
I0
0
`(i, r
0
) di, (1.36)
which is the area below the curve in Fig. 9, whereas the magnetic energy given
by equation 1.26 is the area above the same curve. (Henceforth subscripts of i
and r will be omitted, as the meaning will remain clear aside from the evaluation
of integrals.) Clearly
\
0
}lJ
(i, r) + \
}lJ
(`, r) = i`; (1.37)
i.e., these two areas add to the area of the rectangle with upper corner at the
point (i, `) ; so
\
0
}lJ
(i, r) = i` ÷\
}lJ
(`, r) . (1.38)
The last equation can be reduced to a function of only i and r because ` is a
function of i.
Theorem 16
d\
0
}lJ
(i, r) = `di + ()
}lJ
) dr. (1.39)
Proof. Take the di¤erential of equation 1.38:
d\
0
}lJ
(i, r) = d (i`) ÷d\
}lJ
(`, r)
Note that d (i`) = id` + `di and d\
}lJ
(`, r) = id` ÷ ()
}lJ
) dr (by equation
1.20). Substitution of these last two expressions into the above equation gives
d\
0
}lJ
(i, r) = id` + `di ÷[id` ÷()
}lJ
) dr] = `di + ()
}lJ
) dr.
In general
\
0
}lJ
(i, r) =
_
1
0
d\
0
}lJ
and it follows from equation 1.39 that the result \
0
}lJ
(i, r) will be independent
of the path 1
0
chosen in the i ÷r plane. Note that the result in equation 1.36
1.13. MAGNETIC FORCE FROM COENERGY 21
was obtained by choice of one such path. Hence i and r are state variables for
the coenergy, whereas ` and r were taken as state variables for the energy. The
general reasoning process here is analogous to that in Section 1.9. The choice
to use energy or coenergy is primarily based on mathematical convenience in
the context of a given problem.
Magnetic coenergy is also obtained by integration of the …eld energy density
over the volume \ of the …eld:
\
0
}lJ
=
_
\
_
_
1
0
1 dH
0
_
d\ (1.40)
for soft magnetic material (where H = 0 gives 1 = 0). With constant perme-
ability j (i.e., 1 = jH), this gives
\
0
}lJ
=
_
\
_
jH
2
2
_
d\. (1.41)
Equation 1.40 must be modi…ed to
\
0
}lJ
=
_
\
_
_
1
1c
1 dH
0
_
d\, (1.42)
whereas equation 1.27 remains unchanged, for hard magnetic material (where
H = H
c
gives 1 = 0) as in permanent magnets. This follows because energy
and coenergy are zero when 1 = 0 which is equivalent to H = H
c
. These results
follow from …eld theory. Equation 1.42 is general because it includes equation
1.40 as the special case (soft material) where H
c
= 0.
1.13 Magnetic Force from Coenergy
\
0
}lJ
(i, r) is already established as a function of the two (state) variables i and
r, hence
d\
0
}lJ
(i, r) =
__
0\
0
}lJ
(i, r)
0i
_
r=constant
_
di +
__
0\
0
}lJ
(i, r)
0r
_
I=constant
_
dr.
(1.43)
Matching coe¢cients of di and dr in this equation with those in equation 1.39,
we obtain
` =
_
0\
0
}lJ
(i, r)
0i
_
r=constant
(1.44)
and
)
}lJ
=
_
0\
0
}lJ
(i, r)
0r
_
I=constant
. (1.45)
Equation 1.44 provides the means to deduce the ‡ux ` from the coenergy func-
tion \
0
}lJ
(i, r). Equation 1.45 provides the means to deduce mechanical force
22 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
)
}lJ
from the coenergy function \
0
}lJ
(i, r) and can be converted to a function
of ` and r by insertion of the function relating ` to i. The reasoning here is
analogous to that in Section 1.10 where more detail can be found.
1.14 Torque from Coenergy
Consider a system with rotational rather than linear mechanical motion. The
derivation of magnetic …eld coenergy in this rotational system is analogous to
that in Section 1.12 for systems with linear displacement. Now torque t
}lJ
and
angular displacement 0 replace force )
}lJ
and linear displacement r, respectively,
and likewise become the mechanical terminal variables in Fig. 3. So magnetic
coenergy becomes
\
0
}lJ
(i
0
, 0
0
)
.
=
_
I0
0
`(i, 0
0
) di (1.46)
in place of equation 1.36 and likewise
\
0
}lJ
(i, 0) = i` ÷\
}lJ
(`, 0) (1.47)
replaces equation 1.38. Then analogous to Theorem 16 we have
Theorem 17
d\
0
}lJ
(i, 0) = `di + (t
}lJ
) d0.
Proof. Take the di¤erential of equation 1.47:
d\
0
}lJ
(i, 0) = d (i`) ÷d\
}lJ
(`, 0)
Note that d (i`) = id` + `di and d\
}lJ
(`, 0) = id` ÷ (t
}lJ
) d0 (by equation
1.32). Substitution of these last two expressions into the above equation gives
d\
0
}lJ
(i, r) = id` + `di ÷[id` ÷(t
}lJ
) d0] = `di + (t
}lJ
) d0.
By analogy to Section 1.13 we obtain
i =
_
0\
0
}lJ
(i, 0)
0i
_
0=constant
(1.48)
and
t
}lJ
=
_
0\
0
}lJ
(i, 0)
00
_
I=constant
(1.49)
in place of equations 1.44 and 1.45, respectively.
1.14. TORQUE FROM COENERGY 23
Example 18 Any Linear Magnetic System
Here we repeat Example 10 by using coenergy instead of energy.
By de…nition of a linear system
` = [1(r)] i.
In this case the coenergy is, by application of equation 1.36:
\
0
}lJ
(i
0
, r)
.
=
_
I0
0
`(i, r) di =
_
I0
0
[1(r)] idi =
_
1
2
_
[1(r)] i
2
0
.
The force (same result as obtained in Example 10) is, by equation 1.45:
)
}lJ
=
_
0\
0
}lJ
(i, r)
0r
_
I=constant
=
_
0
_
1
2
_
[1(r)] i
2
0r
_
I=constant
=
_
i
2
2
_ _
d1(r)
dr
_
.
Example 19 (Example 3.5- Fitzgerald et al)
For the relay of Example 3.2, …nd the force on the plunger if the coil is driven
by a controller which produces a current
i (r) = 1
0
_
r
d
_
A.
Note that both the current i (r) and the force are functions of r. Based on the
assumptions in Example 3.2 (which is Example 8 and Fig. 6 here) this is a
magnetically linear system and the following result was obtained:
1(r) =
j
0
·
2
|d
_
1 ÷
r
J
_
2q
.
By application of the result from Example 18
)
}lJ
=
_
i
2
2
_ _
d1(r)
dr
_
= ÷
_
i
2
2
_ _
j
0
·
2
|
2q
_
.
Force as a function of r is obtained by substitution of i (r) into the previous
equation:
) )
}lJ
= ÷
_
1
2
0
j
0
·
2
|
4q
_
_
r
d
_
2
.
Also from Example 18 the coenergy here is
\
0
}lJ
(i, r) =
_
i
2
2
_
[1(r)] =
_
i
2
2
_
_
j
0
·
2
|d
_
1 ÷
r
J
_
2q
_
.
which can be written correctly as a function of r alone by substitution of i (r) :
\
0
}lJ
(i, r) =
_
1
2
0
j
0
·
2
|d
_
1 ÷
r
J
_
4q
_
_
r
d
_
2
.
24 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
However direct application of Equation 1.45 to this last expression would give
an incorrect result )
}lJ
(di¤erent from that found above) because partial dif-
ferentiation with respect to r imposes the condition that the current i be held
constant (at least mathematically). I.e., this condition has been violated already
by the substitution i (r) which renders i a function of r in the last expression.
1.15. ENERGY AND COENERGY CONSIDERATIONS 25
1.15 Energy and Coenergy Considerations
1.15.1 Equivalence of Forces
Equations 1.31 and 1.45 must yield the same result for the force )
}lJ
when
applied to the same device. These equations are repeated here for reference as
)
}lJ
= ÷
_
0\
}lJ
(`, r)
0r
_
X=constant
.
= ÷
_
lim
4r!0
´\
}lJ
(`, r)
´r
_
X=constant
(1.50)
and
)
}lJ
=
_
0\
0
}lJ
(i, r)
0r
_
I=constant
.
=
_
lim
4r!0
´\
0
}lJ
(i, r)
´r
_
I=constant
, (1.51)
respectively, where the right-hand expressions have been added for utility. This
equivalence of forces can be explained by consideration of the `÷i characteristics
of a typical device, as shown in Fig. 10. Note that a family of such characteristics
can be generated by variation of the position parameter r.
The device is assumed to be operating initially along the ` ÷i curve shown
and labelled for a particular position r. Then the position is changed from r
to r + ´r, which moves the device operation onto the curve labelled r + ´r.
The area enclosed between these two curves includes both [÷´\
}lJ
(`
0
, r)[ and
´\
0
}lJ
(i
0
, r), in view of the de…nitions of \
}lJ
and \
0
}lJ
. It is necessary only
to show
÷
_
lim
4r!0
´\
}lJ
(`, r)
´r
_
X=constant
=
_
lim
4r!0
´\
0
}lJ
(i, r)
´r
_
I=constant
; (1.52)
i.e., the right-hand expressions in equations 1.50 and 1.51 converge to the same
quantity as ´r ÷÷0.
The equality in equation 1.52 follows by consideration of the conditions under
which the partial derivatives are de…ned. For equation 1.50 the limit is taken
while ` = `
0
is constant and i is allowed to vary from i
0
to i
0
÷ ´i, which
corresponds to bounding an area (hence bounding [÷´\
}lJ
(`
0
, r)[) in Fig. 10
by the two curves and the horizontal line from point a to point /. For equation
1.51 the limit is taken while i = i
0
is constant and ` is allowed to vary from `
0
to
`
0
÷´`, which corresponds to bounding an area (hence bounding ´\
0
}lJ
(i
0
, r))
in Fig. 10 by the two curves and the vertical line from point a to point c. These
two areas (hence [÷´\
}lJ
(`
0
, r)[ and ´\
0
}lJ
(i
0
, r)) di¤er by only the amount
in the (roughly) triangular region bounded by the line a/, line ac, and curve
segment /c. The ratio of this di¤erence amount to the remaining large bounded
area goes to zero as ´r ÷÷0. The establishes the claimed equality.
As a general rule, it will be more convenient to use energy to determine force
or torque if ‡ux is given as a driving or regulated quantity. Likewise it will be
more convenient to use coenergy to determine force or torque if current is given
as a driving quantity. The latter case is quite common, as shown by subsequent
examples.
26 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.15.2 Direction of Forces
The …eld force )
}lJ
acts to decrease the magnetic energy at constant ‡ux or
increase the magnetic coenergy at constant current. This follows directly by
interpretation of equations 1.50 and 1.51, respectively. In the case of a single
electrical terminal-pair, increase of coenergy at constant current implies that
the force acts to increase the inductance of a device. I.e., the force pulls device
components in a way which would reduce the reluctance in the corresponding
magnetic-circuit model.
1.15.3 Linear Systems
In a linear system ` = [1(r)] i, from which it follows by previous results that
magnetic energy and coenergy are equal:
`
2
21
=
1i
2
2
. (1.53)
Likewise it can be shown from …eld theory that energy and coenergy densities
are equal:
1
2
2j
=
jH
2
2
. (1.54)
In a nonlinear system ` is not linearly proportional to i and equivalently
1 is not linearly proportional to H. Hence in that case the above relations do
not hold. Also, examination of Fig. 10 shows that in general the energy and
coenergy in a nonlinear system are not equal.
1.16 General Solutions
A general solution for the magnetic …eld is necessary in systems where an equiv-
alent magnetic circuit is inadequate or innaccurate. E.g., such systems may
include complex geometry and/or very nonlinear magnetic materials. General
solutions are obtained by numerical algorithms implemented in commercial soft-
ware ("…eld solvers"). The …nite-element method is a common example of such
algorithms.
The energy or coenergy method is still usable even if a general …eld solu-
tion is required. I.e., in this case the energy and coenergy can be computed
by application of equation 1.27 and equation 1.40 or 1.42, respectively. There-
upon forces are calculated by taking partial derivatives (as before), where such
derivatives must be determined by numerical approximation. E.g., commercial
software is available to determine magnetic coenergy for a linear-displacement
actuator as a function of displacement r, from which force can be determined
as described above.
1.16. GENERAL SOLUTIONS 27
Example 20 (Example 3.6- Fitzgerald et al)
See Fig. 11 (RE: Fitzgerald- p. 135, Fig. 3.12 )
Magnetic circuit with stator and rotor free to turn about axis is shown.
Magnetic materials (stator and rotor): permeability is assumed in…nite (j ÷÷
·).
Neglect e¤ects of fringe …elds.
/
.
= axial length (height perpendicular to page); q
.
= air-gap length; r
1
.
of rotor.
Dimensions: / = 1.8 mm; q = 3.0 mm; r
1
= 2.5 cm.
MMF (·i) Data: To be determined in solution (i.e., · and i are not speci-
…ed).
==a) Derive an expression for torque acting on rotor, in terms of dimensions
and magnetic …eld in the two air gaps.
== b) Assume maximum ‡ux density 1 = 1.65 T (to avoid excessive satu-
ration of steel) in portion of air gap where rotor and stator overlap. Compute
maximum torque.
a) In the (overlapping) air gap:
H

=
·i
2q
because the net air gap length is 2q (due to two gaps in series), 1
s|ttl
= 1

is
…nite, and H
s|ttl
=
1
steel
µ
= 0 due to j ÷÷·.
Coenergy:
\
0
}lJ
=
_
\
0
}lJ
_

+
_
\
0
}lJ
_
s|ttl
where
_
\
0
}lJ
_

and
_
\
0
}lJ
_
s|ttl
are found by integration of coenergy density
(equation 1.41) over the volumes of the air gap and the steel, respectively.
_
\
0
}lJ
_
s|ttl
= 0
because
µ1
2
steel
2
=
1
2
steel

= 0 in view of H
s|ttl
= 0 (above). This leaves
\
0
}lJ
=
_
\
0
}lJ
_

=
_
\
_
j
0
H
2

2
_
d\ =
_
j
0
H
2

2
_
\
where the integral is the application of equation 1.41 to the (overlapping) air
gap. The integral reduces to the simple product of coenergy density
_
µ
0
1
2
ag
2
_
and volume \ because H

is taken as constant in the (overlapping) air gap.
By simple geometry
\ = 2q/(r
1
+ 0.5q) 0
28 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
closely approximates the volume of a cylindrical slice where radius r
1
+ 0.5q is
taken midway in the gap and 0 is given in radians.
) \
0
}lJ
=
_
j
0
H
2

2
_
\ =
_
j
0
H
2

2
_
(2q/(r
1
+ 0.5q) 0) =
j
0
(·i)
2
/(r
1
+ 0.5q) 0
4q
Torque follows by application of equation 1.49:
t
}lJ
=
_
0\
0
}lJ
(i, 0)
00
_
I=constant
=
_
_
0
_
µ
0
(ÞI)
2
|(:1+0.5¸)0

_
00
_
_
I=constant
=
j
0
(·i)
2
/(r
1
+ 0.5q)
4q
.
This is a constant torque which is present only when there is some overlap (i.e.,
when 0 ,= 0). Its sign is positive so that it acts in a direction to increase the
overlap angle 0, which tends to align the rotor and stator pole faces.
b) It is necessary to determine mmf (·i) to apply the above equation for
torque. First …nd H

: 1

= j
0
H

==
H

=
1

j
0
=
1.65
4¬ [10
7
]
= 1.31
_
10
6
¸
A, m.
Then from …rst equation in part a):
·i = 2qH

= 2
_
3
_
10
3
¸_ _
1.31
_
10
6
¸_
= 7860 A÷turns.
Now evaluate above equation for torque:
t
}lJ
=

_
10
7
¸
(7860)
2
_
1.8
_
10
2
¸_ _
2.5
_
10
2
¸
+ 0.5
_
3
_
10
3
¸__
4 (3 [10
3
])
= 3.09 N m.
N.B.: Refer to Practice Problem 3.6 (Fitzgerald et al) to determine torque
t
}lJ
by …rst deriving inductance 1. Then t
}lJ
follows by application of t
}lJ
=
_
I
2
2
_ _
JJ(0)
J0
_
.
1.17. MULTIPLY-EXCITED MAGNETIC FIELD SYSTEMS 29
1.17 Multiply-Excited Magnetic Field Systems
1.17.1 Background
Electromechanical systems with multiple electrical terminals are quite common.
E.g., for power (j = ·i) measurement by a wattmeter it is necessary to obtain
torque proportional to the product of two electrical variables (voltage · and
current i). Likewise most energy conversion devices require multiple electrical
terminals to excite multiple windings which produce the magnetic …elds for
intended operation.
1.17.2 Methods of Analysis
The energy and coenergy methods will be extended and applied here to systems
with two electrical terminal-pairs, where analogous extension to more than two
terminal pairs is also possible. This amounts to a generalization of the approach
presented so far. A conceptual representation of a system with two electrical
terminal-pairs and one mechanical terminal-pair is shown in Fig. 12. It is
assumed that the mechanical portion of the system in Fig. 12 undergoes rotary
motion.
The system state is determined by three independent variables because there
are three terminal-pairs. Angular displacement 0 is usually chosen as the me-
chanical state variable. Either ‡ux linkages `
1
and `
2
or currents i
1
and i
2
may
be chosen as electrical state variables. A hybrid set is also possible.
1.17.3 Analysis by Energy Method
The di¤erential energy is
d\
}lJ
(`
1
, `
2
, 0) = i
1
d`
1
+ i
2
d`
2
÷(t
}lJ
) d0, (1.55)
for a system with two electrical terminal-pairs and one mechanical terminal-
pair, analogous to equation 1.32. In this case `
1
, `
2
, and 0 are the natural
choice for state variables because they are the "running variables" to determine
\
}lJ
(`
1
, `
2
, 0) by integration of equation 1.55. The currents are given by
i
1
=
_
0\
}lJ
(`
1
, `
2
, 0)
0`
1
_
0=constant
X2=constant
(1.56)
and
i
2
=
_
0\
}lJ
(`
1
, `
2
, 0)
0`
2
_
0=constant
X1=constant
, (1.57)
analogous to equation 1.30. The torque is given by
t
}lJ
= ÷
_
0\
}lJ
(`
1
, `
2
, 0)
00
_
X1=constant
X2=constant
(1.58)
30 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
analogous to equation 1.35.
The energy is
\
}lJ
(`
10
, `
20
, 0
0
) =
_
X2
0
0
i
2
(`
1
= 0, `
2
, 0 = 0
0
) d`
2
+
_
X1
0
0
i
1
(`
1
, `
2
= `
20
, 0 = 0
0
) d`
1
(1.59)
by integration of equation 1.55 over the speci…c path shown in Fig. 13. Anal-
ogous to the path in Fig. 4 for the singly-excited case, this path begins by in-
tegrating over 0 while `
1
= `
2
= 0 which contributes nothing because t
}lJ
= 0
when there is no magnetic …eld. Then this path completes by integrating over
`
2
while `
1
= 0 and …nally integrating over `
1
. The quantity \
}lJ
(`
10
, `
20
, 0
0
)
found by integration over any other path will be the same as that computed by
equation 1.59 because energy is a state function.
1.17.4 Analysis by Coenergy Method
The coenergy for a system with two electrical terminal-pairs is
\
0
}lJ
(i
1
, i
2
, 0)
.
= `
1
i
1
+ `
2
i
2
÷\
}lJ
(`
1
, `
2
, 0) , (1.60)
analogous to equation 1.38 for a system with one electrical terminal-pair.
Theorem 21
d\
0
}lJ
(i
1
, i
2
, 0) = `
1
di
1
+ `
2
di
2
+ (t
}lJ
) d0. (1.61)
Proof. Take the di¤erential of equation 1.60:
d\
0
}lJ
(i
1
, i
2
, 0) = d (`
1
di
1
) + d (`
2
di
2
) ÷d\
}lJ
(`
1
, `
2
, 0)
Note that d (i
1
`
1
) = i
1
d`
1
+`
1
di
1
, d (i
2
`
2
) = i
2
d`
2
+`
2
di
2
, and d\
}lJ
(`
1
, `
2
, 0) =
i
1
d`
1
+ i
2
d`
2
÷ (t
}lJ
) d0 (by equation 1.55). Substitution of these last three
expressions into the above equation gives
d\
0
}lJ
(i
1
, i
2
, 0) = i
1
d`
1
+ `
1
di
1
+ i
2
d`
2
+ `
2
di
2
÷[i
1
d`
1
+ i
2
d`
2
÷(t
}lJ
) d0]
= `
1
di
1
+ `
2
di
2
+ (t
}lJ
) d0.
Remark 22 This theorem is analogous to Theorem 17.
It follows from equation 1.61, by derivation analogous to that for equations
1.48 and 1.49, that
`
1
=
_
0\
0
}lJ
(i
1
, i
2
, 0)
0i
1
_
0=constant
I2=constant
, (1.62)
1.17. MULTIPLY-EXCITED MAGNETIC FIELD SYSTEMS 31
`
2
=
_
0\
0
}lJ
(i
1
, i
2
, 0)
0i
2
_
0=constant
I1=constant
, (1.63)
and
t
}lJ
= ÷
_
0\
0
}lJ
(i
1
, i
2
, 0)
00
_
I1=constant
I2=constant
. (1.64)
Note that t
}lJ
is determined directly in terms of currents- a signi…cant advan-
tage.
The coenergy is
\
0
}lJ
(i
10
, i
20
, 0
0
) =
_
I2
0
0
`
2
(i
1
= 0, i
2
, 0 = 0
0
) di
2
+
_
I1
0
0
`
1
(i
1
, i
2
= 0, 0 = 0
0
) di
1
(1.65)
by integration of equation 1.61. The path of integration is analogous to that
used to derive equation 1.59.
1.17.5 Linear Systems
For a linear system
`
1
= 1
11
i
1
+ 1
12
i
2
(1.66)
and
`
2
= 1
21
i
1
+ 1
22
i
2
, (1.67)
where 1
11
, 1
22
are self-inductances, 1
12
, 1
21
are mutual inductances, and all
inductances are functions of angular position 0 (in general). Further 1
21
= 1
12
due to the physical reciprocity of magnetic-…eld phenomena.
1.17.6 Application of Energy Method to Linear Systems
The immediate goal is to determine \
}lJ
(`
10
, `
20
, 0
0
) for a linear system in
terms of `
10
, `
20
, 0
0
and the given inductance parameters 1
11
, 1
22
, 1
12
, 1
21
, by
application of equation 1.59. For this it is necessary to express i
1
and i
2
as
functions of `
1
and `
2
which are the running variables of integration in equation
1.59. Hence we …nd i
1
and i
2
by inversion of equations 1.66 and 1.67:
i
1
=
1
22
`
1
÷1
12
`
2
1
(1.68)
and
i
2
=
÷1
21
`
1
+ 1
11
`
2
1
(1.69)
where
1 = 1
11
1
22
÷1
12
1
21
. (1.70)
32 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
Then evaluation of equation 1.59, after substitution of equations 1.68, 1.69, and
1.70, gives
\
}lJ
(`
10
, `
20
, 0
0
) =
_
X2
0
0
[1
11
(0
0
)] `
2
1(0
0
)
d`
2
+
_
X1
0
0
[1
22
(0
0
)] `
1
÷[1
12
(0
0
)] `
20
1(0
0
)
d`
1
(1.71)
=
[1
11
(0
0
)] `
2
20
21(0
0
)
+
[1
22
(0
0
)] `
2
10
21(0
0
)
÷
[1
12
(0
0
)] `
10
`
20
1(0
0
)
.
The torque can be determined by application of equation 1.58, which re-
quires partial di¤erentiation of the above expression with respect to 0
0
. Un-
fortunately partial di¤erentiation of this somewhat complicated expression for
\
}lJ
(`
10
, `
20
, 0
0
) will generate a very complicated expression for the torque,
in particular due to functions of 0
0
which are present in all numerators and
denominators. It turns out that the coenergy method leads to much simpler
expression of results.
1.17.7 Application of Coenergy Method to Linear Sys-
tems
The immediate goal is to determine \
0
}lJ
(i
10
, i
20
, 0
0
) for a linear system in terms
of i
10
, i
20
, 0
0
and the given inductance parameters 1
11
, 1
22
, 1
12
, 1
21
, by appli-
cation of equation 1.65. Evaluation of equation 1.65, after direct substitution
of equations 1.66 and 1.67, yields
\
0
}lJ
(i
1
, i
2
, 0) =
[1
11
(0)] i
2
1
2
+
[1
22
(0)] i
2
2
2
+ [1
12
(0
0
)] i
1
i
2
. (1.72)
The torque is
t
}lJ
= ÷
_
0\
0
}lJ
(i
1
, i
2
, 0)
00
_
I1=constant
I2=constant
=
_
i
2
1
2
_
d [1
11
(0)]
d0
+
_
i
2
2
2
_
d [1
22
(0)]
d0
+[i
1
i
2
]
d [1
12
(0)]
d0
(1.73)
by direct application of equation 1.64 to equation 1.72. Clearly the coenergy
method is far more e¢cient than the energy method in the case of linear systems
and gives the torque directly in terms of currents.
Example 23 (Example 3.7- Fitzgerald et al)
See Figs. 14, 15 (RE: Fitzgerald- pp. 140-41, Figs. 3.15, 3.16 )
Magnetic circuit with stator and rotor free to turn about axis is shown.
System has two electrical terminal-pairs (rotor and stator windings) for ex-
citation.
Inductances are speci…ed directly:
1
11
= (3 + cos 20)
_
10
3
¸
;
1
22
= 30 + 10 cos 20;
1
12
= 0.3 cos 0.
1.18. SYSTEMS WITH PERMANENT MAGNETS 33
Excitation Currents are: i
1
= 0.8 A; i
2
= 0.01 A.
== Find and plot torque t
}lJ
(0) .
Solution 24 Torque can be computed from equation 1.73.:
t
}lJ
=
_
i
2
1
2
_
d [1
11
(0)]
d0
+
_
i
2
2
2
_
d [1
22
(0)]
d0
+ [i
1
i
2
]
d [1
12
(0)]
d0
=
_
i
2
1
2
_
_
÷2
_
10
3
¸
sin20
_
+
_
i
2
2
2
_
(÷20 sin20) ÷[i
1
i
2
] (0.3) sin0
= ÷1.64
_
10
3
¸
sin20 ÷2.4
_
10
3
¸
sin0.
Remark 25 The torque components
_
i
2
1
2
_
d [1
11
(0)]
d0
=
_
i
2
1
2
_
_
÷2
_
10
3
¸
sin20
_
and
_
i
2
2
2
_
d [1
22
(0)]
d0
=
_
i
2
2
2
_
(÷20 sin20)
are due to winding currents acting separately as in a singly excited system.
These torques occur because self inductances are a function of rotor position.
So they are dependent on the reluctance of the magnetic path where both indi-
vidual winding currents actually "see" the same magnetic path. They are called
reluctance torques and act in a direction to maximize coenergy, hence maximize
self inductances. I.e., these torques tend to align the rotor axis with the stator
axis.
The torque component
[i
1
i
2
]
d [1
12
(0)]
d0
= ÷[i
1
i
2
] (0.3) sin0
is due to the mutual interaction between rotor and stator currents and is hence
called the mutual interaction torque. This is also dependent on the reluctance
of the magnetic path and acts in a direction to maximize mutual inductance.
It acts in a direction to align the rotor and stator, hence align their magnetic
…elds.
1.18 Systems with Permanent Magnets
1.18.1 Background
Here we consider an electromechanical system or device which includes a perma-
nent magnet. For simplicity we assume that the system as given has no electrical
terminal-pairs but note that our method can be generalized to systems which
include multiple electrical terminal-pairs.
34 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
We wish to apply the energy/coenergy method so need to determine same.
In the most di¢cult cases a general …eld solution will yield \
}lJ
or \
0
}lJ
by
application of equation 1.27 or equation 1.42, respectively. Either of these equa-
tions will integrate over volume the magnetic …eld created in the system by the
permanent magnet. Note that in the absence of any electrical terminal pair
\
}lJ
and \
0
}lJ
will be functions of only displacement r and the system (in-
cluding magnet) parameters. Then magnetic force )
}lJ
can be determined by
application of equation 1.31 to \
}lJ
or equation 1.45 to \
0
}lJ
. The absence of
an electrical terminal pair actually simpli…es the partial derivatives in these last
equations.
1.18.2 Coenergy Method with Magnetic Circuit
We wish to retain the advantages that magnetic-circuit analysis provides in
those systems which are su¢ciently uniform or simple to permit its application
without excessive error. It turns out that a "…ctitious" or auxiliary winding
with an electrical terminal current expedites the analysis here, in which case
current is the natural choice for a state variable. Hence we choose the coenergy
method, for which current is the natural state variable and for which analysis
of linear systems is most e¢cient (as shown previously).
The …rst task is to determine \
0
}lJ
as a function of displacement r, where we
consider the simple system in Fig. 16 for de…niteness. Previously this was done
by integration of equation 1.39 but so far there is no state variable i. Now there
is only a state variable )
}lJ
to be integrated with respect to r, but )
}lJ
is actually
the desired result. In addition we must account for the e¤ect of the permanent
magnet in determination of the …eld by magnetic-circuit analysis. These issues
are resolved by introduction of the …ctitious winding (a mathematical arti…ce)
shown in Fig. 16. Initially the winding current i
}
a value 1
}0
which cancels the magnetic …eld in the permanent magnet (hence
reduces the …eld to zero in the entire system). At this point we procede to the
integration of equation 1.39 where the second state variable has become i = i
}
,
with ` = `
}
.
The integration is performed over the path 1
1
in the i
}
÷r plane, as shown in
Fig. 17. This path goes from the point (i
}
= 1
}0
, r = 0) where the system …eld
(hence \
0
}lJ
) is zero to the point (i
}
= 0, r = r
0
) where i
}
= 0 is equivalent to
removing the …ctitious winding. I.e., we seek \
0
}lJ
(i
}
= 0, r = r
0
) by moving
on the path 1
1
between these two points through the (deliberately chosen)
segments 1
1o
and 1
1b
.
) \
0
}lJ
(i
}
= 0, r = r
0
) =
_
11a
d\
0
}lJ
+
_
1
1b
d\
0
}lJ
. (1.74)
Note that
_
11a
d\
0
}lJ
= 0 because `
}
= 0 (as well as di
}
= 0) due to the null
…eld condition. Hence
\
0
}lJ
(i
}
= 0, r = r
0
) =
_
1
1b
d\
0
}lJ
=
_
0
1
f0
`
}
(i
}
, r = r
0
) di
}
(1.75)
1.18. SYSTEMS WITH PERMANENT MAGNETS 35
where the term )
}lJ
dr does not appear here because dr = 0 on path 1
1b
. (Note
that the development here was analogous to that which led from equation 1.24
to equation 1.26, but here the "o¤set" current 1
}0
was necessary to render the
integral zero along the …rst path segment.)
The result in equation 1.75 is completely general so it applies in cases where
any magnetic material in the system (including the magnet itself) is nonlinear.
The magnetic force )
}lJ
due to the permanent magnet alone follows by applica-
tion of equation 1.45 to \
0
}lJ
(i
}
= 0, r) , where we note that neither \
0
}lJ
nor
)
}lJ
is zero.
Example 26 (Example 3.8- Fitzgerald et al)
See Fig. 18 (RE: Fitzgerald- p. 145, Fig. 3.19 )
Magnetic circuit including permanent magnet (samarium-cobalt) with mov-
able plunger is shown.
MMF Data: Fictitious winding (·
}
turns with current i
}
) is included for
analytic purposes.
Core and plunger materials: permeability is assumed in…nite (j ÷÷ ·);
neglect …eld fringing e¤ects
r
.
= plunger displacement (movable gap); q
0
.
= …xed air-gap length; d =
magnet length; 1 = depth of core and plunger/
Dimensions: d = 2.0 cm; q
0
= 0.2 cm; 1 = 3.0 cm.;
Widths: \
n
= 2.0 cm; \
¸
= 3.0 cm; \
0
= 2.0 cm.
== a) Derive coenergy \
0
}lJ
as a function of plunger position r.
== b) Derive force on plunger as a function of plunger position r.
== c) Evaluate force at r = 0 and r = 0.5 cm.
Solution 27 a) Represent the DC magnetization curve (RE: Fitzgerald- p. 36,
Fig. 1.19) of samarium-cobalt by the linear approximation
1
n
= j
1
_
H
n
÷H
0
c
_
= j
1
H
n
+ 1
:
where 1
:
= ÷j
1
H
0
c
with j
1
= 1.05j
0
, H
0
c
= ÷712 kA/m,[1
:
[ = 0.94T. Note
that the apparent coercivity H
0
c
adjusted for the slight downward bend in the
1 ÷H curve is somewhat larger than the actual coercivity.
The other ‡ux densities are
1
¸
= j
0
H
¸
in the movable gap and
1
0
= j
0
H
0
in the …xed gap.
Flux continuity requires
1
n
\
n
1 = 1
¸
\
¸
1 = 1
0
\
0
1
36 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
and by Ampere’s law (for mmf):
·
}
i
}
= H
n
d + H
¸
r + H
0
q
0
.
There are six unknowns 1
n
, 1
¸
, 1
0
, H
n
, H
¸
, H
0
and six equations (where
the ‡ux continuity equation counts as two). In matrix terms the system is
sparse so it is simplest to reduce to a three-by-three system by combining some
equations. I.e.,
j
1
_
H
n
÷H
0
c
_
\
n
= j
0
H
¸
\
¸
by substitution of the …rst and second equations into the …rst "half" of the
fourth equation. In a similar vein,
j
0
H
¸
\
¸
= j
0
H
0
\
0
by substitution of the second and third equations into the last "half" of the
fourth equation. The last three equations form the three-by-three system which
can be written in matrix form:
_
_
j
1
\
n
÷j
0
\
¸
0
0 j
0
\
¸
÷j
0
\
0
d r q
0
_
_
_
_
H
n
H
¸
H
0
_
_
=
_
_
j
1
H
0
c
\
n
0
·
}
i
}
_
_
.
This is solved for the vector of unknowns H
n
, H
¸
, H
0
by matrix inversion and
H
n
is substituted into the …rst equation to yield:
1
n
=
j
1
_
·
}
i
}
÷H
0
c
d
_
d + \
n
_
µ
R
µ
0
__
r
Vg
+
¸0
V0
_.
`
}
= ·
}
\
n
11
n
= [·
}
\
n
1]
_
_
j
1
_
·
}
i
}
÷H
0
c
d
_
d + \
n
_
µ
R
µ
0
__
r
Vg
+
¸0
V0
_
_
_
where `
}
= 0 when
i
}
= 1
}0
=
H
0
c
d
·
}
i
}
=
÷1
:
d
j
1
·
}
.
Then the coenergy is
\
0
}lJ
(i
}
= 0, r = r
0
) =
_
0
H
0
c
d
N
f
i
f

}
\
n
1]
_
_
j
1
_
·
}
i
}
÷H
0
c
d
_
d + \
n
_
µ
R
µ
0
__
r
Vg
+
¸0
V0
_
_
_
di
}
=
_
_
\
n
1(1
:
d)
2
2j
1
_
d + \
n
_
µ
R
µ
0
__
r
Vg
+
¸0
V0
__
_
_
,
1.18. SYSTEMS WITH PERMANENT MAGNETS 37
by application of equation 1.75.
b) The force is
)
}lJ
= ÷
_
¸
_
\
2
n
1(1
:
d)
2
2j
0
\
¸
_
d + \
n
_
µ
R
µ
0
__
r
Vg
+
¸0
V0
__
2
_
¸
_,
by application of equation 1.45 to \
0
}lJ
(i
}
= 0, r = r
0
) , where the negative sign
indicates that the force acts to decrease the movable gap by pulling the plunger
inward.
c) Evaluation of the above expression for )
}lJ
by substitution of known
parameters gives
)
}lJ
= ÷115 N at r = 0 cm
and
)
}lJ
= ÷85.8 N at r = 0.5 cm.
38 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.18.3 Equivalent of Magnet with Linear Material
Claim 28 Consider a section of linear, hard magnetic material:
1
n
= j
1
_
H
n
÷H
0
c
_
(1.76)
with area ¹ and length d. With respect to the external magnetic circuit which it
faces, this section can be replaced by:
1. a section of soft, linear magnetic material with the same permeability j
1
(1 = j
1
H) and same geometry;
2. and an equivalent winding of
·i = ÷H
0
c
d ampere-turns. (1.77)
I.e., this will result in same ‡ux in the external magnetic circuit.
Proof. By Ampere’s law
H
n
d +T
t
= 0 (1.78)
where T
t
represents the mmf developed at the "terminals" of the external mag-
netic circuit which the magnet faces.
) H
n
=
÷T
t
d
. (1.79)
The ‡ux produced in the external magnetic circuit by the permanent magnet is
= ¹1
n
= ¹
_
j
1
_
H
n
÷H
0
c
__
1m=
Fe
d
= j
1
¹
_
÷H
0
c
÷
T
t
d
_
.
Now consider the replacement:
= j
1
¹
_
·i
d
÷
T
t
d
_
by magnetic-circuit analysis with application of linear superposition. This equa-
tion gives the same result as the previous equation because
ÞI
J
= ÷H
0
c
by
equation 1.77 for the mmf of the equivalent winding.
Remark 29 Note the analogy between the previous theorem for magnetic cir-
cuits and Thevenin’s theorem for electrical networks.
1.18. SYSTEMS WITH PERMANENT MAGNETS 39
Example 30 (Example 3.9- Fitzgerald et al)
See Fig. 19 (RE: Fitzgerald- p. 149, Fig. 3.22 )
Magnetic circuit including permanent magnet (neodymium-iron-boron) with
movable plunger is shown.
MMF Data: Excitation winding (·
1
= 1500 turns with current i
1
) is in-
cluded.
Core and plunger materials: permeability is assumed in…nite (j ÷÷ ·);
neglect …eld fringing e¤ects
r
.
= plunger displacement (movable gap); q
0
.
= …xed air-gap length; d =
magnet length; 1 = depth of core and plunger/
Dimensions: d = 8 mm; q
0
= 1 mm; 1 = 3.5 cm.
Widths: \ = 4.0 cm; \
1
= 4.5 cm.
== a) Find r÷ directed force on plunger when current i
1
= 0.
== b) Find current i
1
required to reduce plunger force to zero.
Solution 31 a) Represent the DC magnetization curve (RE: Fitzgerald- p. 36,
Fig. 1.19) of neodymium-iron-boron by the linear approximation
1 = j
1
_
H ÷H
0
c
_
= j
1
H + 1
:
where 1
:
= ÷j
1
H
0
c
with j
1
= 1.05j
0
, H
0
c
= ÷940 kA/m,[1
:
[ = 1.25T. Apply
Claim 28 to replace magnet by section of linear material with permeability j
1
and equivalent winding of
·i = ÷H
0
c
d = ÷
_
÷9.4
_
10
5
¸_ _
8
_
10
3
¸_
= 7520 ampere-turns.
The equivalent magnetic circuit after this replacement is shown in Fig. 19b,
where the two mmf sources are in series with the variable gap, …xed gap, and
magnet reluctances
¹
r
=
r
j
0
\
1
1
,
¹
0
=
q
0
j
0
\1
,
and
¹
n
=
d
j
1
\1
,
respectively. For i
1
= 0 this is equivalent to a single-winding system driven by
(·i)
tjuIu
where
\
0
}lJ
=
1i
2
2
=
(·i)
2
tjuIu
2 [¹
r

0

n
]
.
The force on the plunger is
)
}lJ
=
_
0\
0
}lJ
0r
_
Iequiv=constant
=
(·i)
2
tjuIu

r

0

n
]
2
_

r
dr
_
=
(·i)
2
tjuIu
(j
0
\
1
1) [¹
r

0

n
]
2
40 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
which gives
)
}lJ
= ÷703 N
upon substitution of known parameters.
b) This occurs by setting the net mmf to zero:
(·i)
tjuIu
+ ·
1
i
1
= 0
which gives
i
1
=
(·i)
tjuIu
·
1
=
7520
1500
= 5.01 A.
The direction of the current cannot be determined because the direction of
magnetization of the magnet is not given. This current must be applied in the
direction which reduces net ‡ux to zero.
1.18. SYSTEMS WITH PERMANENT MAGNETS 41
42 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
1.18. SYSTEMS WITH PERMANENT MAGNETS 43
44 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION
44
Chapter 2
Rotating Machines
2.1 Introduction
Certain basic principles are common to both AC and DC rotating machines,
whereby a physical machine can be reduced to a simple mathematical model.
The key principle is electromagnetic induction (Faraday’s law):
c =
d`
dt
(2.1)
where
` = ÷·c (2.2)
and
c
.
=
_
S
1 do. (2.3)
Electromechanical energy conversion occurs when mechanical motion produces
JX
J|
,= 0. In rotating machines there are three methods to obtain
JX
J|
,= 0 me-
chanically:
1. Rotation of windings through a 1 …eld;
2. Rotation of a 1 …eld past windings;
3. Variation of reluctance in the main magnetic ‡ux path by rotor rotation,
where rotor has geometry designed for that purpose.
` varies periodically by any of these methods due to the cyclic nature of
mechanical motion in a rotating machine.
2.2 General Rotating Machine
The general machine (a generator or a motor) consists of a stator (stationary
member) and a rotor (rotating member) where the stator partially or totally
45
46 CHAPTER 2. ROTATING MACHINES
encloses the rotor, as shown in Fig. 1. Stator and rotor cores usually consist
of magnetic material to concentrate and shape the 1 …elds produced by their
respective windings.
An armature winding is a set of interconnected coils through which the bulk
of machine energy passes, whether energy is electrical output (generator) or
electrical input (motor). The …eld winding produces the operating ‡ux of the
machine and usually carries DC current. It may be replaced by permanent
magnets in some machines.
Torque is produced by the interaction of stator and rotor magnetic …elds
and acts to align them. In a generator mechanical torque is applied and the
aforementioned torque opposes rotation. In a motor electrical energy (armature
current) is applied and the aforementioned torque aids rotation.
2.2.1 AC Machines
The armature winding is on the stator and the …eld winding is on the rotor
almost invariably. This corresponds to method 2 in Section 2.1 to produce
JX
J|
,= 0.
Synchronous Machine
The …eld winding is excited by DC current and requires rotating electrical con-
tacts (slip rings) to feed it.
Induction Machine
The …eld winding is shorted and no external source is applied directly to it so
no rotating electrical contacts are required. Rather AC induction excites it by
generalized transformer action between the stator and rotor, as explained later.
Variable Reluctance Machine
There is no …eld winding. A varying 1 …eld is produced by a non-uniform
air-gap reluctance with respect to rotor angle. Time-varying armature (stator)
currents are applied.
2.2.2 DC Machine
The armature winding is on the rotor and the …eld winding is on the stator in
most cases, which is a reversal of positions with respect to an AC machine. This
corresponds to method 1 in Section 2.1 to produce
JX
J|
,= 0. The DC interface
between excitation or output and the periodic ("AC") induced armature (speed)
voltage is performed by a commutator on the rotor. The commutator switches
current in synchronism with rotation; it is the underlying reason for placing
the armature on the rotor. An exception is the so-called brushless DC motor
in which the rotor is a permanent magnet. The armature of this machine is
2.3. STATOR AND ROTOR CORES 47
on the stator and driven by electronic switching circuitry which emulates a
commutator.
2.3 Stator and Rotor Cores
Material for these cores is usually magnetically soft and highly permeable elec-
trical steel, with windings placed in core slots. This core material increases
coupling between windings and the 1 …eld strength (hence stored magnetic en-
ergy) in a machine. It also allows the designer to shape and distribute the 1
…elds. Core losses due to hysteresis and eddy currents must be considered, so
laminated construction is used to reduce eddy currents. Eddy currents may
occur in both stator and rotor of variable reluctance machines. Refer to Figs.
4.1, 4.2, 4.3 in textbook.
2.4 Some Machine Type-Names
1. DC
2. Synchronous
3. Permanent magnet (PM)
4. Induction
5. Variable Reluctance
6. Hysteresis
7. Brushless
2.5 Similarity of Physical Principles
The following machines are apparently quite di¤erent but share a common prin-
ciple with regard to the interaction between stator and rotor …elds.
2.5.1 DC Machine
Both the stator and rotor 1 …elds are …xed in space, hence at a …xed angu-
lar displacement with respect to each other. The latter …eld is …xed due to
commutator action. Torque is produced by the tendency of these …elds to align.
2.5.2 AC Induction Machine
Both the stator and rotor 1 …elds rotate together in space and remain at a …xed
angular displacement with respect to each other. Actually the stator …eld can
be analyzed as a superposition of forward and backward rotating waves, where
48 CHAPTER 2. ROTATING MACHINES
the backward wave is zero in a 3-phase machine but the same as the forward
wave in a single-phase machine. Torque is produced by the tendency of these
…elds to align, just as in a DC motor.
2.6 Di¤erentiation Rule for Integrals (Leibnitz)
Leibnitz rule is stated here for reference because it will be used to apply Fara-
day’s law to ‡ux integrals. Given
1 (.) =
_
b(:)
o(:)
) (r, .) dr (2.4)
where both the integrand ) (r, .) and the limits a (.) and / (.) may be functions
of a parameter .. Also, in general the integrand ) (r, .) may contain another
integral. We have
d1 (.)
d.
=
_
d/ (.)
d.
_
) (/ (.) , .) dr ÷
_
da (.)
d.
_
) (a (.) , .) +
_
b(:)
o(:)
0) (r, .)
0.
dr
(2.5)
which is essentially a superposition of e¤ects due to changes in the upper limit,
lower limit, and integrand with respect to changes in a parameter .. This is
restated for evaluation at a speci…c value . = .
0
to show that the derivative
operator can be moved to precede the integral in the last term:
_
d1 (.)
d.
_
:=:0
=
_
d/ (.)
d.
_
:=:0
) (/ (.
0
) , .
0
) dr÷
_
da (.)
d.
_
:=:0
) (a (.
0
) , .
0
)+
0
0.
_
b(:0)
o(:0)
) (r, .) dr.
(2.6)
2.7 Gap Field Distribution in Rotating Machines
Clearly a gap between the rotor and the stator is necessary to permit rotor
rotation. The 1 …eld produced by the …eld winding continues through the
gap and into armature slots where it is accessible to the armature winding.
Coincidentally a prescribed distribution of the 1 …eld in the gap, with respect
to the axis of the …eld winding, is necessary to achieve the desired output voltage
waveform in a generator or desired torque characteristic in a motor. It turns out
that nearly all magnetic …eld energy is concentrated in the gap volume because
the H …eld is very small in highly permeable cores. For a given energy in the
gap volume, the distributed nature of the 1 …eld reduces the tendency toward
magnetic saturation at points in the cores where 1 attains its extreme values.
In other words, the magnetization is spread to attain a more or less uniform
degree of magnetic-dipole alignment throughout the core volume. Analytically
this magnetization can be represented by equivalent surface currents on the
cores, as previously described.
2.8. MULTIPOLE FIELD DISTRIBUTION 49
The 1 …eld is normal to the rotor and stator surfaces due to boundary
conditions on the magnetic core materials. Hence 1 is directed radially in the
case of a uniform circular gap. For analytic purposes we assume that the 1
…eld is essentially constant across the gap from rotor to stator if the gap is
small relative to the radius of the rotor. Further we assume that this 1 …eld
may vary with angle 0 (in cylindrical coordinates) around the circumference of
the gap but does not vary with linear (axial) displacement along the rotor or
stator. These principles apply to a 1 …eld irrespective of whether it is produced
by current in a …eld winding or an armature winding.
2.8 Multipole Field Distribution
Consider the radially directed spatial 1 …eld around the circumference of the
gap due to an instantaneous current in either a …eld winding or an armature
winding. At any position on the circumference speci…ed by angle 0, the …eld 1 is
speci…ed by its magnitude and sense (i.e., radially directed outward or inward).
Note that around the circumference 1 undergoes more or less complete reversals
in direction because a winding sets up circulating currents. The number of poles

¡olts
) in 1 is by de…nition the number of maxima which occur in [1[ during
a complete (360

) revolution on the circumference of the gap. ·
¡olts
is always
an even number because no magnetic monopoles exist.
·
¡olts
is determined only by the winding pattern if the gap is uniform (e.g.,
round rotor and round stator with distributed winding). ·
¡olts
will be depen-
dent primarily on the design of the pole faces if the gap is non-uniform (e.g.,
salient-pole rotor or stator with concentrated winding). Refer to Figs. 4.4, 4.5,
4.6, and 4.7 for illustrations of two-pole and four-pole machines including spatial
distributions of their 1 …elds. In a machine with a large number of pole-pairs
each pair "spans" a small arc and contributes a small or narrow segment of the
1 …eld in a prescribed manner, by design of the winding distribution or pole
face geometry. Generally one-half subcycle of the 1 …eld occurs in each of these
narrow segments.
In most multipole machines, by design the magnetic …elds of the stator and
rotor will have the same number of poles to achieve optimal performance. Hence
the machine output, whether it be voltage in a generator or torque in a motor,
can be taken as the aggregate of the outputs with respect to all the magnetic
poles.
An analogy can be drawn with bar magnets and horseshoe magnets. A
simple bar magnet has two poles hence represents a two-pole winding. Bar
magnets can be curved into horseshoe magnets to form a four-pole or eight-pole
machine as shown in Fig. 3.
50 CHAPTER 2. ROTATING MACHINES
2.9 Induced Voltage by Application of Faraday’s
Law
_
c
1 d: = ÷
d
dt
_
S
1 da = ÷
dc
dt
(2.7)
where o may be chosen as any surface for which the curve C is a boundary and
the total ‡ux is
c
.
=
_
S
1 da. (2.8)
Hence the induced voltage in path C is
c =
_
c
1 d: = ÷
dc
dt
and the ‡ux linkage to an ·÷turn winding on this path is
`
.
= ÷·c (2.9)
so the induced voltage in the winding is
c = ÷·
dc
dt
=
d`
dt
. (2.10)
We apply equation 2.10 to a rotor (armature) winding consisting of a single
rectangular loop (· = 1) turning through a 1 …eld produced in the gap by the
stator (…eld) winding, as shown in Fig. 2. This corresponds to method 1 in
Section 2.1 to produce
JX
J|
,= 0. Here the "curve" C is taken as the rectangular
loop. The surface o is taken as the cylindrical half-shell whose base is the planar
section enclosed by the rectangular loop and whose cyclindrically curved surface
coincides with the circular path of the loop conductors around the gap. The
curve C and this surface o are shown in Fig. 2. This choice of o expedites
evaluation of the surface integral in equation 2.8 because the gap 1 …eld is
everywhere normal to this surface. (By conservation of ‡ux, the same result
would be obtained for the surface integral by choosing o as the aforementioned
planar section but its evaluation would be complex.)
The plane of the rectangular loop is oriented at an arbitrary angle 0 with
respect to the vertical as shown in Fig. 2. Thus its upper and lower wires bound
the cylindrical half-shell at angles 0 and 0 +¬, respectively. This gives
c
.
=
_
S
1 da =
0+t
_
0
1
_
0
0
, t
_
|d0
0
, (2.11)
2.10. OPTIMAL GAP FIELD 51
where | is the length of the loop, for the integral over the surface of the cylindrical
half-shell. Application of Leibnitz rule from equation 2.6, where we now take
. = t, yields
c (t
0
) = ÷
_
dc
dt
_
|=|0
= ÷|
_
_
_
_
_
_
_
_
d
dt
0(|0)+t
_
0(|0)
1
_
0
0
, t
_
d0
0
. ¸¸ .
transformer voltage term
+ [1(0 (t
0
) +¬) ÷1(0 (t
0
))]
_
d0
dt
_
|=|0
. ¸¸ .
speed voltage term
_
_
_
_
_
_
_
_
.
The speed voltage can be expressed as
c
s¡ttJ
(t
0
) = ÷| [1(0 (t
0
) +¬) ÷1(0 (t
0
))] . (t
0
) (2.12)
= ÷
_
|
r
_
· [1(0 (t
0
) +¬) ÷1(0 (t
0
))]
. ¸¸ .
Lorentz force density
where .
.
=
J0
J|
=
u
:
. Note that the Lorentz force density term appears in equation
2.12 because this produces the 1 …elds down the length of the wires. By design
of machine windings the 1 …eld is normally anti-symmetric around the gap:
1(0 (t
0
) +¬) = ÷1(0 (t
0
)) , (2.13)
so
c
s¡ttJ
(t
0
) = | [21(0 (t
0
))] . (t
0
) (2.14)
and …nally if . is constant
c
s¡ttJ
(t) = .| [21(.t)] . (2.15)
Thus we conclude that the waveform c
s¡ttJ
(t) is simply a scaled version of
1(0) where 0 = .t; for a generator this can be taken as the output voltage.
This coincides with the classic "cutting of ‡ux" interpretation; see Example 4.4
in textbook.
Analogously we can apply equation 2.10 to a stator (armature) winding
consisting of a single rectangular loop (· = 1) where a rotating 1 …eld is
produced in the gap by the rotor (…eld) winding. This corresponds to method
2 in Section 2.1 to produce
JX
J|
,= 0. Now the speed voltage c
s¡ttJ
(t) induced
in the stator can be determined simply by taking the rotating rotor as the
spatial frame of reference. Hence the …eld appears stationary in time and can
be denoted by 1(0), with 0 relative to the rotor axis. Now the stator winding
appears to be rotating with respect to the rotor …eld and equation 2.15 applies
directly.
2.10 Optimal Gap Field
Optimal operation of a rotating machine requires that the (Fourier) spectral
content of the induced speed voltage waveform c
s¡ttJ
(t) matches the spectral
52 CHAPTER 2. ROTATING MACHINES
content of the required armature (generator) output or applied armature (mo-
tor) input. In an AC machine this specializes to the simple requirement that
c
s¡ttJ
(t) is a pure sinusoid; i.e., its spectrum contains no harmonics. In a DC
machine this must be interpreted by taking the intervening commutator into
account, so the armature itself actually sees or should provide a square wave.
By equation 2.15, the gap …eld waveform 1(0) has the same spectral content
as that required in c
s¡ttJ
(t). So 1(0) should be a pure (fundamental) sinusoid
in an AC machine or a square wave (i.e., piecewise constant) in a DC machine.
The importance of the spectral content in c
s¡ttJ
(t) hence 1(0) is explained
as follows. Assume that the machine is in normal steady-state operation, ro-
tating at a constant angular velocity .
n
. Consider the electrical circuit or loop
formed by the armature winding and a voltage source of the required spectral
content connected directly to the armature terminals. By Kircho¤’s voltage
law (KVL), any mismatch in spectral content between c
s¡ttJ
(t) and the voltage
source yields a voltage di¤erence which contains additional harmonic compo-
nents with respect to .
n
. These harmonic components in the loop voltage
will drive corresponding current components around the loop. These current
components are undesirable because they increase the RMS current value hence
increase resistive (i
2
1) losses and increase core eddy current losses (which are
proportional to the square of frequency) but do not contribute to average electro-
mechanical power transfer. The last statement follows by application of the or-
thogonality principle to harmonics with respect to the fundamental component.
In addition harmonics tend to introduce small periodic variations in .
n
.
2.11 AC Machines
2.11.1 Electrical and Mechanical Frequency
The spatial variation in the gap 1 …eld will be more rapid in machines with a
greater number of poles:
0
ot
=
_
·
¡olts
2
_
0
o
(2.16)
where 0
ot
is the electrical angle or argument of the cyclic …eld 1(0
ot
) and 0
o
is the mechanical or spatial angle around the gap. It follows that
.
t
=
_
·
¡olts
2
_
.
n
, (2.17)
hence
)
t
=
_
·
¡olts
2
_
)
n
, (2.18)
by di¤erentiation of equation 2.16 with respect to time. From this last equation
it follows that
)
t
=
_
·
¡olts
2
__
11'
60
_
Hz, (2.19)
2.11. AC MACHINES 53
where the term
111
60
converts mechanical revolutions per minute to )
n
which
is in revolutions per second. Note that )
t
is the electrical frequency of the speed
voltage and )
n
is the mechanical frequency of rotation in a given machine.
2.11.2 Synchronous Machines
Synchronous machine operation is best explained by separate consideration of
motors and generators. An AC motor is operating synchronously if )
t
, the
electrical frequency of the speed voltage which it produces, equals the electrical
frequency of the voltage or current applied to its armature winding. Then )
n
,
the mechanical frequency of rotation, is constrained or locked to the applied
frequency by equation 2.18. Conversely an AC generator is operating synchro-
nously if )
n
is such that )
t
given by equation 2.18 matches the frequency of the
(possibly "in…nite") bus to which it is connected. Note that the inherent "fre-
quency conversion" given by equation 2.18 always occurs in multipole machines
(both generators and motors).
Example 1 Given a 2-pole single-phase machine. What RPM is required to
obtain )
t
= 60 Hz ?
Solution 2 From the above equation
11' =
_
2
·
¡olts
_
(60)
t
) =
_
2
2
_
(60 60) = 3600.
Note that the spatial B …eld of this machine completes one full cycle per me-
chanical revolution.
Example 3 Given a 4-pole single-phase machine. Express electrical frequency
in terms of mechanical frequency.
Solution 4 ·
¡olts
= 4, so by equation
)
t
=
_
·
¡olts
2
_
)
n
= )
t
=
_
4
2
_
)
n
= 2)
n
2.11.3 Induction Machines
In induction machines the armature (stator) winding is the same as in synchro-
nous machines. On the other hand, the …eld (rotor) winding is short-circuited
so that no external excitation need be applied to it.
AC rotor current follows by induction (generalized transformer action) from
stator to rotor. E.g., consider a two-pole machine (·
¡olts
= 2) where .
n
< .
t
so that the rotor rotation continuously slips behind rotation of the stator 1
…eld. Hence there will be a continuosly induced rotor current of frequency
.
s
= .
t
÷.
n
(2.20)
54 CHAPTER 2. ROTATING MACHINES
due to relative motion between the rotor winding and the rotating armature
…eld. Note that a frequency change and electromechanical energy conversion
occur here.
Induction motors are the most common and widely applied of all motors.
On the other hand induction generators are rarely found except recently in wind
power applications for which they are somewhat suitable.
The squirrel-cage rotor or reasonable facsimile is a common method of con-
struction for induction motors. Here the "windings" are solid aluminum bars
cast in slots and shorted together by common cast aluminum end rings. Conse-
quently induction motors are durable, reliable, and relatively inexpensive, which
are contributing factors to their immense popularity.
Refer to Fig. 4.15 for a torque-speed curve of an induction motor. Note that
the induction motor must operate at less sychronous speed to develop torque.
Otherwise there is no slip, hence no induced rotor current.
2.11.4 Three-Phase Machines
Phase refers to the electrical phase or delay between AC currents in multiple
armature windings. A 3-phase system can be either a synchronous machine
or an induction machine where there are three sets of armature windings per
pole-pair; refer to Fig. 4.12 in textbook. These sets are displaced by 120 electri-
cal degrees in the spatial pattern around the rotor or stator. Hence in general
the mechanical displacement angle between sets is (2¬,3) , (·
¡olts
,2) . E.g., in
a two-pole machine, these sets are symmetrically displaced by 120 mechani-
cal degrees. E.g., in a four-pole machine the two coils for each phase can be
connected in series (or in parallel because coil voltages are identical); see Fig.
4.12b in textbook. Likewise the currents supplied to or voltages generated from
3-phase windings are symmetrically displaced by 120 electrical degrees.
In a three-phase system the electrical frequency is the same as in a single-
phase system. I.e., electrical frequency is determined solely by the number of
poles in a machine. The e¤ect of a balanced three-phase system is to produce a
rotating ‡ux wave which has only a forward component, which will be shown by
phasor addition of rotating waves. Hence three-phase motors possess constant
torque vs. time; i.e., there is no ‡uctuating double-frequency torque component.
Most of the world’s power systems (hence most synchronous generators) are 3-
phase. The attendant lack of ‡uctuating "reaction torque" makes it possible to
build and operate extremely large 3-phase generators which would otherwise be
damaged by such ‡uctuating torques.
2.11.5 Design Considerations
An AC machine (generator or motor) requires a large number of poles (·
¡olts
)
by equation 2.19 if it operates at a relatively low mechanical RPM . (Electrical
power systems operate at )
t
= 60 Hz or )
t
= 50 Hz in North America or Eu-
rope, respectively.) Salient-pole rotors are most suitable in this case from the
standpoint of mechanical construction and economy. Rotor windings must be
2.12. DC MACHINES 55
concentrated so the 1 …eld will be determined only by the rotor (gap) geometry
and not by a winding distribution. E.g., consider a large hydroelectric generator
where the prime mover is water power; Fig. 4.1 shows such a generator where
mechanical rotation is at 37 RPM. E.g., in a salient-pole motor more space is
available to accomodate rotor windings.
An AC machine (generator or motor) requires a small number of poles

¡olts
) by equation 2.19 if it operates at a relatively high mechanical RPM.
Round rotors (hence a uniform gap) are a necessity in this case because salient
poles would incur unacceptably high stress and windage loss. Rotor windings
are distributed so the 1 …eld will be determined only by the distribution and
not by the gap. E.g., consider a large generator where the prime mover is a
gas or steam turbine operating at 1800 or 3600 RPM; Fig. 4.11 shows such a
generator. In this case ·
¡olts
= 2 or ·
¡olts
= 1, respectively.
2.12 DC Machines
A DC machine may be viewed as a "commutated" version of a synchronous AC
machine. I.e., all rotating machines execute cyclic mechanical motion which
creates a cyclic (AC) induced speed voltage in the windings to support electro-
mechanical energy conversion. Such speed voltage has zero average (DC) value
so a form of recti…cation is required to achieve a non-zero DC value. In a con-
ventional DC machine this recti…cation is attained by placing a commutator and
hence the armature on the rotor shaft while the (DC-excited) …eld winding is
on the stator. The commutator mechanically "switches" the connection of the
armature coils to external terminals in synchronism with machine rotation, by
means of copper segments and carbon brushes. Fig. 4.17 shows an elementary
DC machine with a commutator.
The instantaneous induced speed voltage in a single turn of the rotor winding
is proportional to the gap 1 …eld at the turn’s particular angular position 0,
as shown previously. All turns between two diametrically opposed commutator
segments are e¤ectively in series, so it is possible to deduce the output winding
voltage as a function of time from the 1 …eld distribution. It turns out that the
ideal stator 1 …eld is piecewise constant (a square wave) to maximize the DC
component of the commutated (recti…ed) speed voltage.
The spatial orientation between the rotor and stator …elds is nearly constant
during machine operation. I.e., the stator …eld is inherently …xed whereas the
rotor …eld is stationary with respect to the active positive and negative com-
mutator segments. This allows the rotor …eld to revolve in space by only the
arc subtended by the active commutator segments, which is small if segments
are numerous. Typically by design (brush positions, etc.) the stator and rotor
…elds are perpendicular to each other to maximize torque.
56 CHAPTER 2. ROTATING MACHINES
2.13 Design of Gap Field
2.13.1 Magnetic Circuit Analysis
Assume that a certain gap …eld 1(0) is required in the design of a given machine.
The …eld 1(0) varies periodically with 0 due to continuous rotation around
the gap, where the spatial electrical frequency is
Þ
poles
2
electrical cycles per
mechanical revolution. It will be shown below by magnetic-circuit analysis that
two options are available to design or "shape" 1(0) :
1. Variation of the gap length q (0) versus 0 by pole-face geometry;
2. Winding distribution versus 0 to produce a prescribed driving mmf func-
tion T (0).
Only the …rst option is feasible with respect to salient poles where the wind-
ings must be concentrated. Only the second option is feasible with respect to
non-salient poles where the gap must be uniform (e.g., round rotor AC machine).
The following magnetic-circuit analysis is applicable to 1(0) which is pro-
duced by either a rotor or stator winding. This analysis is illustrated in Fig.
4 and Fig. 5 for the non-salient (uniform gap) and salient (non-uniform gap)
cases, respectively, where a rotor winding is shown for de…niteness. Note that
the magnetic path C shown in each …gure passes through the origin and is
symmetric with respect to it. Application of Ampere’s law to path C yields
_
c
H d| = T (0) (2.21)
with
T (0) =
_
S
J da (2.22)
where J is the current density through surface o. Equation 2.21 reduces to
_
1
j
0
_
[1(0) q (0) ÷1(0 +¬) q (0 +¬)] = T (0) (2.23)
where 1 = j
0
H in the gap, the reference direction for 1 is everywhere ra-
dially outward, and negligible reluctance is assumed for the stator and rotor
core material. Now we assume anti-symmetry of the current or current-sheet
distribution and symmetry of the gap, which are written as
J (0 +¬) = ÷J (0) (2.24)
and
q (0 +¬) = q (0) , (2.25)
respectively. This implies anti-symmetry of the …eld:
1(0 +¬) = 1(0) . (2.26)
2.13. DESIGN OF GAP FIELD 57
Hence equation 2.23, with subsitution of equations 2.25 and 2.26, gives
_
1
j
0
_
[21(0) q (0)] = T (0) (2.27)
or
1(0) =
j
0
T (0)
2q (0)
=
j
0
T
o¸1
(0)
q (0)
. (2.28)
where
T
o¸1
(0) =
T (0)
2
(2.29)
is the mmf across one gap. Equation 2.28 shows that a required 1 …eld shape can
be obtained by varying q (0) (…rst option) or by varying T (0) (second option),
where in either case the other variable is (at least piecewise) constant.
2.13.2 Idealized Current Density
The idealized current density J which corresponds to a prescribed T (0) follows
from equation 2.22 by di¤erentiation of both sides with respect to 0. In order
to di¤erentiate we render the integral on the right-hand side of equation 2.22
as a function of 0 (without r) in polar coordinates, by assuming that J is
a thin current sheet on the surface of the rotor. E.g., if T (0) = sin0 then
J (0) = ÷cos 0. In the case of a discrete winding, J (0) is the objective to be
approximated by the turns distribution.
2.13.3 MMF of Discrete Windings
We determine the mmf T (0) corresponding to certain discrete winding distri-
butions by direct evaluation of equation 2.22, which is converse to the process
described above to …nd the idealized J (0) from a prescribed T (0) . However the
idealized J (0) still serves as a guide for modi…cation of a winding distribution
to improve T (0) . T (0) is called the spatial mmf wave which so far is stationary
in space.
Example 5 Consider the simplest case where an idealized concentrated rotor
winding lies totally within a very small arc (approaching zero) located at 0 = ±
t
2
.
The rotor may be non-salient or salient, as illustrated in Fig. 4 or Fig. 5,
respectively. Assume that this winding consists of · turns carrying the current
i through the surface o. Here T
o¸1
(0) is simply a piecewise constant function
(square wave) with values ±
ÞI
2
as shown in Fig. 4.19. I.e., an abrupt change
of polarity occurs in the evaluation of equation 2.22 when the linear segment of
path C rotates through 0 = ±
t
2
.
Remark 6 The linear presentation of mmf T
o¸1
(0) and gap q (0) in Fig. 4.19
is called developed form. I.e., these circular functions have been mapped to linear
form or "laid out ‡at".
58 CHAPTER 2. ROTATING MACHINES
Example 7 Consider the armature winding shown in Fig. 4.20, which corre-
sponds to a simpli…ed two-pole three-phase machine, where the air gap is uniform
(ignoring slots). Each phase winding consists of ·
c
turns uniformly distributed
over one third of the stator circumference and wound in two layers to simplify
access to end conductors. Idealizing the phase a winding as a uniform current
sheet J over one third of the circumference, T
o¸1
(0) is a trapezoidal wave with
extreme values ±
ÞcI
2
. A linear transition between these extremes occurs with …-
nite slope due to the evaluation of equation 2.22 when the linear segment of path
C is rotating through the current sheet J. The actual windings are discrete with
24 total slots, where each side of the phase a winding occupies 4 slots. Hence
the linear transition in T
o¸1
(0) is now approximated by 4 steps, as shown in
Fig. 4.20b. Refer to Example 4.1a,b for an extension of this problem.
Example 8 Consider the rotor winding shown in Fig. 4.21, where the number
of turns per slot (and possibly the spacing between slots) is varied. This is done
to obtain an mmf wave T
o¸1
(0) which is a closer aproximation to a sinusoid,
as shown in Fig. 4.21b.
2.13.4 Fourier Analysis of MMF Waves
The spatial mmf wave T
o¸1
(0) is periodic in 0, with period 2¬ for a two-pole
machine. So the standard analytic approach is to represent the developed form
of the mmf wave T
o¸1
(0) by a Fourier series in the variable 0. Note that this
analysis can be generalized to arbitrary ·
¡olts
simply by substitution of 0
t
for
0.
2.14 Gap Field in AC Machines
In an AC machine the optimal gap …eld 1(0) is a pure sinusoid as explained
in Section 2.10. Hence by equation 2.28 the desired mmf wave T
o¸1
(0) is a
pure sinusoid if the machine has a uniform gap q (0). T
o¸1
(0) approaches a
pure sinusoid only if the Fourier coe¢cients of all harmonics approach zero, so
our objective is that only the fundamental coe¢cient be non-zero. Now the
distributed windings in some previous examples are assessed with respect to
this objective.
In Example 5 the spatial wave T
o¸1
(0) is a square wave with Fourier series
representation
T
o¸1
(0) =
_
·i
2
_
_
1

n=0
¹
n
cos (:0)
_
. ¸¸ .
unit square wave
(2.30)
where
¹
n
=
_
4
tn
, : odd
0, : even
_
. (2.31)
2.14. GAP FIELD IN AC MACHINES 59
From the above series the fundamental component is
´
T
o¸1
(0) =
_
·i
2
__
4
¬
_
cos 0 (2.32)
where
max
_
´
T
o¸1
(0)
_
=
_
·i
2
__
4
¬
_
. (2.33)
(Note that
´
T
o¸1
(0) is denoted in the textbook by simply T
o¸1
(0) .)
In Example 7 we consider a di¤erent spatial wave and denote it here by
T
0
o¸1
(0) . Its Fourier series representation is
T
0
o¸1
(0) =
_
·i
2
_
_
¸
_
1

n=0
n odd
¹
0
n
cos (:0)
_
¸
_ (2.34)
where in general
¹
0
n
,= ¹
n
. (2.35)
The waveshape of T
0
o¸1
(0) is a closer approximation to a sinusoid but its fun-
damental component is actually smaller than that of
´
T
o¸1
(0) :
¹
0
1
< ¹
1
. (2.36)
This can be shown from the formula for Fourier coe¢cients, because T
o¸1
(0)
resides at its maximum value (‡at top) for a greater interval in 0. This leads
naturally to the de…nition of the winding factor
/
u
=
¹
0
1
¹
1
(2.37)
for any winding, based on the condition that ·i is the same as in the concen-
trated reference winding whose fundamental coe¢cient is ¹
1
. The fundamental
component here becomes
´
T
0
o¸1
(0) =
_
/
u
·i
2
__
4
¬
_
cos 0, (2.38)
analogous to equation 2.32. The typical range of /
u
is 0.85 to 0.95. It is im-
portant to note that the winding in Example 7 does reduce the strength of
harmonics relative to the fundamental; i.e.,
¸
¸
¸
¸
¹
0
n
¹
0
1
¸
¸
¸
¸
<
¸
¸
¸
¸
¹
n
¹
1
¸
¸
¸
¸
(2.39)
for odd : _ 3. This follows because that winding is a closer approximation to a
sinusoid.
60 CHAPTER 2. ROTATING MACHINES
In a 3-phase system, for each phase equation 2.38 becomes
´
T
0
o¸1
(0) =
_
/
u
·
¡|ost
·
¡olts
_
i
o
_
4
¬
_
cos
__
·
¡olts
2
_
0
o
_
, (2.40)
where
Þ
phase
Þ
poles
is the number of turns per phase (replacing
Þ
2
) and i
o
is the
current per phase. This is equivalent to equation (4.5) in the textbook.
Refer to Example 4.1c,d and Example 4.2.
2.15 Gap Field in DC Machines
In a DC machine the optimal gap …eld distribution 1(0) is a (piecewise con-
stant) square wave as explained in Section 2.10. Hence by equation 2.28 the
desired mmf wave T
o¸1
(0) is also a square wave, which corresponds to a con-
centrated (stator) …eld winding if ideally the machine has a uniform gap q (0).
The ideal uniform gap is only approximated over a large arc in a real machine
due to physical winding and magnetic circuit constraints. Elementary, two-pole,
and four-pole DC machines with concentrated (stator) …eld windings are shown
in Figs. 4.17, 4.22, and 4.24, respectively. It is apparent from these …gures that
the stator must be technically classi…ed as salient-pole even though the objective
is a uniform gap over the largest possible arc with respect to the round rotor.
In this last regard observe that the four-pole machine is more e¤ective than the
two-pole machine in achieving the largest arc of gap uniformity. The actual
"‡at-topped" 1 …eld obtained in an elementary two-pole machine is shown in
Fig. 4.18.
In a DC machine the rotor winding must be distributed uniformly due to
mechanical and operational constraints imposed by the commutator. This is
shown for two-pole and four-pole machines in Figs. 4.22 and 4.24, respectively.
For both machines the driving rotor mmf T
o¸1
(0) is a stepwise approximation
to a sawtooth wave, where the spatial frequency of the sawtooth in the four-
pole machine is twice that of the two-pole machine. The mmf T
o¸1
(0) for these
machines is shown in developed form in Fig. 4.23 and Fig. 4.24b, respectively,
where the latter uses a current-sheet idealization for the rotor winding.
By magnetic circuit analysis, the peak value of the sawtooth mmf wave is
max (T
o¸1
(0)) =
_
C
o
2:·
¡olts
_
i
o
A turns/pole (2.41)
where
C
o
.
= total number of armature-winding conductors
:
.
= number of parallel paths through armature winding
i
o
= armature current, A.
Hence by Fourier analysis of the sawtooth mmf wave, the peak value of its
2.16. GAP FIELD FOR NON-UNIFORM AIR GAP 61
spatial fundamental component is
max
_
´
T
o¸1
(0)
_
=
_
8
¬
2
__
C
o
2:·
¡olts
_
i
o
. (2.42)
2.16 Gap Field for Non-Uniform Air Gap
The variation of air-gap length q (0) is extreme in certain machine designs,
whereas machines under previous consideration possessed air gaps that were
uniform or a least piecewise uniform over a large arc. These extreme cases fall
by de…nition into the category of salient-pole machines.
The analysis which culminated in equation 2.28 may remain valid in the
regions where the gap is small (and usually piecewise uniform). An example is
the salient-pole DC machine shown in Fig. 4.26. This is essentially a reproduc-
tion of Fig. 4.22 which was previously discussed in Section 2.15. However that
analysis does not apply in the region of this machine where the gap is very large
because its simplifying assumptions are usually invalid there. Likewise that
analysis does not apply in a machine where the gap is large everywhere. An
example of this is the salient-pole AC synchronous machine shown in Fig. 4.26b.
Sometimes an analytic solution is still possible in such cases by development of
other simplifying assumptions (later).
In complex cases a general …eld solution is necessary. This can be achieved
by use of numerical methods (e.g., …nite-element analysis). A …nite-element
solution for the magnetic …eld in a salient-pole DC generator is shown in Fig.
4.27.
2.17 Rotating MMF Waves (AC Machines)
2.17.1 Introduction
For a single AC phase, consider the fundamental component of the spatial MMF
wave which is stationary in space while it undergoes a "pulsating" amplitude
variation with respect to time. A decomposition of this wave into the sum of
a forward (travelling) wave and a backward (travelling) wave is possible. A
corresponding decomposition of each phase in a three-phase system is possible,
which yields a forward wave and a backward wave for each phase. In a balanced
three-phase system, the superposition of all forward waves reinforces to produce
a large net forward wave whereas the superposition of all backward waves cancels
to produce a zero net backward wave. These reinforcement and cancellation
properties are shown by phasor analysis to be the result of phase di¤erences or
time delay between the individual-phase MMFs. Note that the analysis here is
also applicable to the harmonics of each phase if necessary.
62 CHAPTER 2. ROTATING MACHINES
2.17.2 Decomposition of Single Phase
For a single phase winding, refer to equation 2.40 for its fundamental spatial
mmf wave (which is stationary in space) and assume the sinusoidal excitation
i
o
= 1
o
cos .
t
t. (2.43)
Dropping the prime notation in
´
T
0
o¸1
(0) , this gives
´
T
o¸1
(0) = 1
max
[cos 0
ot
] [cos .
t
t] (2.44)
where
0
ot
.
=
_
·
¡olts
2
_
0
o
(2.45)
and
1
max
.
=
_
4
¬
__
/
u
·
¡|ost
·
¡olts
_
1
o
. (2.46)
It is understood that
´
T
o¸1
(0) is really
´
T
o¸1
(0, t) in equation 2.44 but the latter
notation is omitted for simplicity. Note that here
´
T
o¸1
(0) remains …xed in space
with the envelope cos 0
ot
while it amplitude is pulsating with respect to time
due to the factor cos .
t
t. Refer to Fig. 4.28.
The decomposition of
´
T
o¸1
(0) into the sum of forward and backward trav-
elling waves is accomplished simply by applying the trig identity
cos ccos , =
_
1
2
_
[cos (c ÷,) + cos (c +,)] (2.47)
to equation 2.44 where we take
c = 0
ot
(2.48)
and
, = .
t
t. (2.49)
This yields
T
o
= T
+
o
+T

o
(2.50)
where the forward-travelling wave is
T
+
o
.
=
_
1
max
2
_
[cos (0
ot
÷.
t
t)] (2.51)
and backward-travelling wave is
T

o
.
=
_
1
max
2
_
[cos (0
ot
+.
t
t)] . (2.52)
Here the subscript a refers to phase a and the superscripts + and ÷ refer to
forward- and backward- travelling waves, respectively. Note that the subscript
aq1 (referring to "air gap 1") and the decoration ´ (referring to fundamental
spatial component) have been dropped to simplify notation but are understood.
This decomposition is illustrated in phasor form by Fig. 4.28c.
2.17. ROTATING MMF WAVES (AC MACHINES) 63
2.17.3 Decomposition of Three-Phase System
Here we simply apply the above result for a single-phase to each phase of a
balanced three-phase system; refer to Fig. 4.29. This requires only modi…cation
of the angle in equation 2.43 to indicate that the currents of phases a, /, and c
are symmetrically displaced in phase by 120

. Hence in phasor notation
1
o
= 1
n
\0

==T
o
= T
+
o
+T

o
, (2.53)
1
b
= 1
n
\ ÷120

==T
b
= T
+
b
+T

b
, (2.54)
and
1
c
= 1
n
\120

==T
c
= T
+
c
+T

c
. (2.55)
The net mmf is one wave
T = T
o
+T
b
+T
c
(2.56)
which can be decomposed into forward- and backward- travelling waves:
T = T
+
+T

. (2.57)
Clearly
T
+
= T
+
o
+T
+
b
+T
+
c
(2.58)
and
T

= T

o
+T

b
+T

c
. (2.59)
The individual forward-wave components are
T
+
o
.
=
_
1
max
2
_
[cos (0
ot
÷.
t
t)] , (2.60)
T
+
b
.
=
_
1
max
2
_
[cos (0
ot
÷120

÷[.
t
t ÷120

])] =
_
1
max
2
_
[cos (0
ot
÷.
t
t)] ,
(2.61)
and
T
+
c
.
=
_
1
max
2
_
[cos (0
ot
÷240

÷[.
t
t ÷240

])] =
_
1
max
2
_
[cos (0
ot
÷.
t
t)] ,
(2.62)
where the cosine arguments 0
ot
, 0
ot
÷ 120

, and 0
ot
÷ 240

are due to the
spatial o¤sets of windings a, /, and c, respectively, and the cosine arguments
.
t
t, .
t
t ÷120

, and .
t
t ÷240

are due to the time (phase) o¤sets of i
o
, i
b
, i
c
,
respectively. Hence
T
+
.
=
_
31
max
2
_
[cos (0
ot
÷.
t
t)] (2.63)
from equation 2.58.
64 CHAPTER 2. ROTATING MACHINES
Likewise, the individual backward-wave components are
T

o
.
=
_
1
max
2
_
[cos (0
ot
+.
t
t)] , (2.64)
T

b
.
=
_
1
max
2
_
[cos (0
ot
÷120

+ [.
t
t ÷120

])] =
_
1
max
2
_
[cos (0
ot
+.
t
t ÷240

)] ,
(2.65)
and
T

c
.
=
_
1
max
2
_
[cos (0
ot
÷240

+ [.
t
t ÷240

])] =
_
1
max
2
_
[cos (0
ot
+.
t
t ÷480

)] (2.66)
=
_
1
max
2
_
[cos (0
ot
+.
t
t ÷120

)]
T

will be found from equation 2.59 where the required summation will be
simpli…ed by using phasor form. (Note that phasor form can only be applied to
groups of waves travelling in the same direction because a reversal of direction
is equivalent to reversal of signs between spatial and temporal arguments in the
cosine function.) The phasor assignments 1
o
, 1
b
, and 1
c
induce a phasor inter-
pretation on 1
max
by means of equation 2.46 interpreted for each phase current.
In turn this induces phasor interpretations on T
+
o
, T
+
b
, T
+
c
and T

o
, T

b
, T

c
through equations 2.51 and 2.52, interpreted for 1
max
of the respective phases.
Hence we can write
¯
T

o
=
_
1
max
2
_
\0

, (2.67)
¯
T

b
=
_
1
max
2
_
\120

(2.68)
and
¯
T

c
=
_
1
max
2
_
\ ÷120

. (2.69)
where the decoration¯indicates the phasor representation of a variable. Applying
equation 2.59 in phasor form:
¯
T

=
_
1
max
2
_
[1\0

+ 1\120

+ 1\ ÷120

] = 0. (2.70)
Hence we conclude that T = T
+
+ T

= T
+
consists of only a forward-
travelling spatial mmf wave whose amplitude is three times that of the forward-
travelling wave due to a single phase acting alone.
Theorem 9 The forward-travelling wave rotates at the synchronous angular
velocity
.
s
.
=
_
2
·
¡olts
_
.
t
. (2.71)
2.18. ANALYSIS OF NONSALIENT-POLE MACHINES 65
Proof. At a given point on the advancing wave
0
ot
÷.
t
t = constant
or
_
·
¡olts
2
_
0
o
÷.
t
t = constant.
For a time interval ´t this gives
_
·
¡olts
2
_
´0
o
÷.
t
´t = 0
or
´0
o
´t
=
.
t
_
Þ
poles
2
_.
This last expression is the actual rate of angular advance and agrees with .
s
.
Refer to Examples 4.3 and 4.5. Refer to Practice Problem 4.4.
2.17.4 Machine Analysis by Flux Waves
The decompositions in the preceding sections can be applied directly to the
analysis of two-pole single-phase and two-pole three-phase synchronous motors
rotating at synchronous speed. The forward mmf wave T
+
produces a 1 …eld
rotating forward which pulls the rotor in synchronism with it, hence producing
a constant forward torque on the rotor winding which carries DC current. The
reverse mmf wave T

produces a 1 …eld rotating backward at 2.
n
with respect
to the rotor hence a double-frequency reverse torque on the rotor winding with
average value zero. The total torque is the superposition of the forward and
reverse torques.
In the single-phase case the torque is net forward but carries the undesirable
double-frequency variation. In the three-phase case the backward torque is zero
because the backward-travelling mmf wave is zero. So in the three-phase case
the net forward torque is absolutely constant.
General AC synchronous machine operation can be characterized in terms
of ‡ux waves as follows. In a generator the rotating …eld of the rotor pulls the
forward-travelling armature ‡ux wave (due to mmf wave) along at synchronous
speed while the electromechanical torque opposes rotation which is sustained
by the prime mover. In a motor the rotating forward-travelling armature ‡ux
wave (due to mmf wave) pulls the rotor along at synchronous speed while the
electromechanical torque aids rotation.
2.18 Analysis of NonSalient-Pole Machines
2.18.1 Introduction
We analyze an idealized elementary nonsalient-pole (uniform air gap) machine
with distributed stator and rotor windings on cores with negligible reluctance.
66 CHAPTER 2. ROTATING MACHINES
This analysis can be easily generalized later to more complex machines. The
machine is treated as an electromechanical system with two electrical terminal-
pairs and one mechanical terminal-pair.
2.18.2 Machine Speci…cation
The derivation applies to a general multipole machine where
0
nt
=
_
·
¡olts
2
_
0
n
, (2.72)
where 0
n
is the angle between the axes of the stator and rotor windings. Refer
to Fig. 4.34 which shows a machine for the special case ·
¡olts
= 2. Optimal
operation is assumed; i.e., the spatial mmf wave
´
T
o¸1
(0) hence the spatial
magnetic ‡ux wave 1(0) are pure (fundamental frequency) sinusoids.
2.18.3 Machine Inductance Parameters
The stator self inductance, rotor self inductance, and rotor-to-stator mutual
inductance are denoted by 1
ss
, 1
::
, and /
s:
(0
nt
) , respectively, where the italic
capital letters denote constant values and the script letter denotes a variable.
Refer to Appendix B.3 (textbook) for the derivation of these inductances with
respect to an assumed spatial fundamental (sinusoidal) mmf wave. It is shown
there that
/
s:
(0
nt
) = 1
s:
cos (0
nt
) (2.73)
where the maximum 1
s:
occurs when the magnetic axes of the stator and rotor
are aligned (0
nt
= 0). The stator and rotor ‡ux linkages are given by:
_
`
s
`
:
_
=
_
1
ss
/
s:
(0
nt
)
/
s:
(0
nt
) 1
::
_ _
i
s
i
:
_
. (2.74)
2.18.4 Network Equations
The terminal voltages are given by:
_
·
s
·
:
_
=
_
1
s
0
0 1
:
_ _
i
s
i
:
_
+
d
dt
_
`
s
`
:
_
(2.75)
where 1
s
and 1
:
are the stator and rotor winding resistances, respectively, and
J
J|
is the time-derivative operator. Substitution of equation 2.74 into equation
2.75 gives
_
·
s
·
:
_
=
_
1
s
0
0 1
:
_ _
i
s
i
:
_
+
d
dt
__
1
ss
/
s:
(0
nt
)
/
s:
(0
nt
) 1
::
_ _
i
s
i
:
__
(2.76)
or (in expanded form)
·
s
= 1
s
i
s
+1
ss
di
s
dt
+
d
dt
[(/
s:
(0
nt
)) i
:
] (2.77)
2.18. ANALYSIS OF NONSALIENT-POLE MACHINES 67
and
·
:
= 1
:
i
:
+1
::
di
:
dt
+
d
dt
[(/
s:
(0
nt
)) i
s
] . (2.78)
Note that in each of these last two equations the terms (from left-to-right) are
resistive (loss), self-induced voltage, and mutual interaction voltage.
Examination of the mutual interaction voltages shows
d
dt
[(/
s:
(0
nt
)) i
:
] =
_
d
dt
(/
s:
(0
nt
))
_
i
:
+ [/
s:
(0
nt
)]
di
:
dt
(2.79)
and
d
dt
[(/
s:
(0
nt
)) i
s
] =
_
d
dt
(/
s:
(0
nt
))
_
i
s
+ [/
s:
(0
nt
)]
di
s
dt
. (2.80)
Note that in each of these last two equations the terms (from left-to-right) are
speed voltage and mutually induced voltage (transformer action). Relevant to
the speed-voltage term we evaluate:
d
dt
(/
s:
(0
nt
)) =
d
dt
(1
s:
cos (0
nt
)) = ÷[1
s:
sin(0
nt
)]
d0
nt
dt
. (2.81)
Finally we have
·
s
= 1
s
i
s
+1
ss
di
s
dt
+ [/
s:
(0
nt
)]
di
:
dt
÷[1
s:
i
:
sin(0
nt
)]
d0
nt
dt
(2.82)
and
·
:
= 1
:
i
:
+1
::
di
:
dt
+ [/
s:
(0
nt
)]
di
s
dt
÷[1
s:
i
s
sin(0
nt
)]
d0
nt
dt
(2.83)
by substitution of equation 2.81 into equations 2.79 and 2.80 followed by sub-
stitution of equations 2.79 and 2.80 into equations 2.77 and 2.78, respectively.
Note that .
nt
.
=
J0me
J|
.
2.18.5 Coenergy by Inductance Parameters
This is a multiply-excited magnetic-…eld system which was previously ana-
lyzed for the general case. Refer to equation 1.72 in the previous notes en-
titled "Electromechanical Energy Conversion" or equivalently equation (3.70)
in the textbook. With reference to that equation, the identi…cation 1
11
(0) =
1
ss
, 1
22
(0) = 1
::
, and 1
12
(0) = /
s:
(0
nt
) yields the coenergy
\
0
}lJ
(i
s
, i
:
, 0
n
) =
1
ss
i
2
s
2
+
1
::
i
2
:
2
+[/
s:
(0
nt
)] i
s
i
:
=
1
ss
i
2
s
2
+
1
::
i
2
:
2
+1
s:
i
s
i
:
cos
__
·
¡olts
2
_
0
n
_
.
(2.84)
This expression for coenergy generalizes to include additional, analogous terms
in the case where there are multiple stator and/or multiple rotor windings.
All self inductances will be constant and mutual inductances will be constant
amongst all windings on one side of the gap (stator or rotor). Only mutual
inductances between windings on opposite sides of the gap will be functions of
0
n
. Note that methods here can be applied in cases where core permeability is
…nite.
68 CHAPTER 2. ROTATING MACHINES
2.18.6 Coenergy by Magnetic Field
The coenergy is found directly here by integration of the average magnetic-…eld
coenergy density over the volume of the air gap. We assume that the magnetic
‡ux density 1(0) in the gap is radial and due to the sum (superposition) of the
stator and rotor spatial fundamental (sinusoidal) mmf waves. These stator and
rotor mmf waves must have the same spatial frequency but there is no restriction
with respect to their relative motion; i.e., either wave may be moving (rotating)
or stationary. Only the magnetic ‡ux which links both stator and rotor coils
contributes to the mutual coenergy in the gap so we are in e¤ect assuming that
leakage ‡ux is negligible. Such leakage ‡ux can be categorized as stator leakage
or rotor leakage including slot, toothtip, end-turn, and space-harmonic leakage.
The peak magnitudes of the stator and rotor mmf waves are denoted by 1
s
and 1
:
, respectively, while the spatial angle between these waves is denoted by
c
s:
. The peak magnitude of the resultant mmf wave (sum of stator and rotor
mmf waves) is denoted by 1
s:
. 1
s:
is the phasor sum of 1
s
and 1
:
where it is
su¢cient to use only the angle c
s:
for the present analysis. Refer to Fig. 4.35
which shows 1
s:
as the diagonal of a parallelogram with sides 1
s
and 1
:
and
included angle c
s:
. Hence by trigonometric formula
1
2
s:
= 1
2
s
+1
2
:
+ 21
s
1
:
cos c
s:
. (2.85)
The coenergy density in the gap is
0\
0
}lJ
0\
=
_
j
0
2
_
H
2
=
_
j
0
2
_
_
1
j
0
_
2
=
1
2
2j
0
, (2.86)
where \ denotes volume, 1 and H are in the gap and functions of 0, and we
have made the substitution H =
1
µ
0
. Substitution of
1 =
j
0
T
o¸1
(0)
q
, (2.87)
which follows from equation 2.28, into the previous equation gives
0\
0
}lJ
0\
=
_
µ
0
Fag1(0)
¸
_
2
2j
0
=
_
j
0
2
_
_
T
o¸1
(0)
q
_
2
, (2.88)
where T
o¸1
(0) is the sinusoidal mmf wave which is the sum of the stator and
rotor mmf waves. Previously we have denoted the peak magnitude of this wave
by 1
s:
. For integration over the gap volume it is su¢cient to use the average
value of
JV
0
fld
J\
over rotation around the gap from 0 = 0 to 0 = 2¬ :
Average
_
0\
0
}lJ
0\
_
= Average
_
_
j
0
2
_
_
T
o¸1
(0)
q
_
2
_
(2.89)
=
_
j
0
2q
2
_
_
Average
_
[T
o¸1
(0)]
2
__
=
_
j
0
2q
2
__
1
2
s:
2
_
=
_
j
0
4
_
_
1
s:
q
_
2
,
2.18. ANALYSIS OF NONSALIENT-POLE MACHINES 69
where the last line follows because the average of the square of a sinusoid is
one-half of the square of its peak value.
Finally, integration over the gap volume:
\
0
}lJ
=
_
Average
_
0\
0
}lJ
0\
__
\ =
_
_
j
0
4
_
_
1
s:
q
_
2
_
[(¬1|) q] , (2.90)
where | and 1 are the axial length and mean diameter of the air gap, respec-
tively, so \
~
= (¬1|) q. This can be rewritten as
\
0
}lJ
=
_
j
0
¬1|
4q
_
1
2
s:
=
_
j
0
¬1|
4q
_
_
1
2
s
+1
2
:
+ 21
s
1
:
cos c
s:
¸
(2.91)
where the …nal expression follows by substitution of equation 2.85. Note that
methods here can be applied in cases where core permeability is …nite.
2.18.7 Torque from Coenergy
In general the torque is given by
T = ÷
_
0\
0
}lJ
(i
s
, i
:
, 0
n
)
00
n
_
Is=constant
Ir=constant
(2.92)
which is equation 1.73 in the aforementioned notes, where the identi…cations
i
1
= i
s
,i
2
= i
:
, and 0 = 0
n
have been made. The torque will be determined by
applying this equation to the two di¤erent coenergy expressions which resulted
by the inductance method and the magnetic-…eld method.
Regarding the inductance method, we apply equation 2.92 to the coenergy
result in equation 2.84. This gives
T = ÷
_
·
¡olts
2
_
1
s:
i
s
i
:
sin
__
·
¡olts
2
_
0
n
_
, (2.93)
where the negative sign indicates that the torque T acts to decrease 0
n
hence
align the axes of the stator and rotor windings. In the case of multiple stator
and/or multiple rotor windings, as described in the comments following equa-
tion 2.84, the above torque equation would include additional, analogous terms
corresponding to all mutual inductances "1
s:
" between windings on opposite
sides of the gap. The torques would still act to align the magnetic axes of
corresponding windings. Refer to Examples 4.6 and 4.7.
Regarding the magnetic-…eld method, we apply equation 2.92 to the coenergy
result in equation 2.91. This gives
T = ÷
_
j
0
¬1|
2q
_
1
s
1
:
sinc
s:
(2.94)
70 CHAPTER 2. ROTATING MACHINES
for the assumed two-pole case, which generalizes to
T = ÷
_
·
¡olts
2
_ _
j
0
¬1|
2q
_
1
s
1
:
sinc
s:
(2.95)
for the multipole case. This generalization can be obtained by superposition of
all mmf waves, where an orthogonality condition is operative between the mmf
waves of di¤erent poles during the integration of magnetic-…eld energy density.
Note that the terms 1
2
s
and 1
2
:
contributed nothing to the partial derivative
because the condition that i
s
and i
:
be constant is equivalent to constant 1
s
and 1
:
. Analogous to the result by the inductance method, the negative sign
indicates that the torque T acts to decrease c
s:
hence align the axes of the
stator and rotor windings. It follows that positive and negative values of c
s:
correspond to generator and motor action, respectively.
Equivalence of the torque equations 2.93 and 2.95 can be shown by invoking
results for 1
s:
from Appendix B.3 (textbook) and previous magnetic circuit
analysis for 1
s
and 1
:
.
The torque given by equation 2.95 can be interpreted as the product of
quadrature components of the stator and rotor mmf waves, by associating the
term sinc
s:
with either 1
s
or 1
:
in this product. Alternative expressions (for
application in appropriate circumstances) are
T = ÷
_
·
¡olts
2
_ _
j
0
¬1|
2q
_
1
s
1
s:
sinc
s
(2.96)
and
T = ÷
_
·
¡olts
2
_ _
j
0
¬1|
2q
_
1
:
1
s:
sinc
:
, (2.97)
where the angles c
s
and c
:
are taken between components corresponding to
stator and resultant and corresponding to rotor and resultant, respectively, as
shown in Fig. 4.35.
Another useful expression for torque is
T = ÷
_
·
¡olts
2
_ _
¬1|
2
_
1
s:
1
:
sinc
:
, (2.98)
which follows from equation 2.97 by substitution of 1
s:
=
¸1sr
µ
0
. Here resultant
‡ux density 1
s:
and mmf 1
:
appear explicity so this idealized result can be
used to determine maximum torque for a given size of machine. I.e., 1
s:
is
limited to about 1.5 to 2.0 T by magnetic saturation of armature teeth and 1
:
is limited by constraints on rotor winding current related to temperature rise,
etc. Hence given speci…c limits on 1
s:
and 1
:
while ·
¡olts
is …xed, increased
torque requires increased area (¬1|) of the gap (hence increased surface area of
the rotor and stator), by equation 2.98.
Equation 2.98 can be rewritten as
T = ÷
_
¬
2
_
_
·
¡olts
2
_
2

s:
1
:
sinc
:
, (2.99)
2.19. SYSTEM ANALYSIS 71
where
s:
is the total ‡ux resultant over the area of one pole. I.e., we have

s:
= [average ¦[1[¦] ¹
¡
(2.100)
where 1 is the magnetic …eld (‡ux density) resultant from combined stator and
rotor mmfs, the average ¦[1[¦ is taken over one-half wavelength, and
¹
¡
.
=
¬1|
·
¡olts
(2.101)
is the surface area of one pole. Now
average ¦[1[¦ =
_
2
¬
_
1
s:
(2.102)
where 1
s:
is the peak value of the sinusoid 1, so

s:
=
__
2
¬
_
1
s:
_ _
¬1|
·
¡olts
_
=
_
21|
·
¡olts
_
1
s:
(2.103)
or
1
s:
=
_
·
¡olts
2
_ _

s:
1|
_
. (2.104)
Finally substitution of this last expression into equation 2.98 gives the result in
equation 2.99.
Refer to Example 4.8 and Practice Problem 4.6.
2.19 System Analysis
2.19.1 Elementary Machine Model
The elementary non-salient pole machine is by assumption a lossless magnetic
energy storage system or device in the preceding analysis; such device analysis
was originally introduced in the notes entitled Electromechanical Energy Con-
version. Here the resulting system equations 2.82, 2.83, and 2.93 specify the
relationships between the variables at the terminal-pairs (ports) of the system.
In this elementary machine, there are two electrical ports and one mechani-
cal port, while at each port there is one "through" variable and one "across"
variable. The "across" variabes are ·
s
and ·
:
at the electrical ports and T
(torque) at the mechanical port. The corresponding "through" variables are
i
s
, i
:
, and 0
nt
, respectively. In each of the three system equations a di¤erent
"across" variable is a function of all the "through" variables. Note that these
system equations and the alternative expression for torque in equation 2.95 are
expressed in terms of instantaneous values (and time derivatives) of i
s
, i
:
, 0
nt
,
and instantaneous values of 1
s
, 1
:
, c
s:
, respectively. The only underlying as-
sumption in their derivation was a uniform-gap machine. Hence these equations
are in general applicable to any type of machine, whether it be AC synchronous,
AC induction, or even DC.
72 CHAPTER 2. ROTATING MACHINES
These nonlinear di¤erential equations govern system behavior, where ad-
ditional constraints between port variables can be imposed by externally con-
nected electrical and mechanical elements. Complete analytical or closed-form
solutions of these equations are generally not possible so numerical methods are
appropriate. However in some cases these system equations can be specialized
(e.g., linearized about an operating point) to expedite a local analytic solution.
2.19.2 Practical Machine Model
For a basic practical machine model we assume that the rotor moment of inertia
J
1
and the rotor and stator winding resistances 1
:
and 1
s
to the elementary model. The moment of inertia gives
d.
nt
dt
=
_
·
¡olts
2
_ _
d0
dt
_
=
_
·
¡olts
2
_ _
T
J
1
_
= ÷
_
·
¡olts
2
_
2
_
1
s:
i
s
i
:
J
1
_
sin(0
nt
)
(2.105)
by application of equation 2.93. The winding resistances were included already
in the electrical-port equations 2.82 and 2.83.
Now this model is generalized slightly to cover the common cases of motors
and generators by adding two more elements. The additional elements are: a
resistance 1
tj
in series with or in place of the applied stator voltage source ·
s
;
and an external torque T
tr|
on the rotor moment of inertia.
1
tj
plays the role of either a source or a load resistance according to the
intended operation of the machine as either a motor or a generator, respectively.
In either case, it can be incorporated into the system equations simply by the
replacement (1
tj
+1
s
) ÷÷1
s
in equation 2.82. For a generator we may take
the applied stator voltage source as zero and use ·
s
now to denote the voltage
tj
.
T
tr|
plays the role of either an opposing load torque or an applied prime-
mover torque according to the intended operation of the machine as either a
motor or a generator, respectively. In either case, it can be incorporated into
the system equations simply by the replacement (T
tr|
+T) ÷÷ T in equation
2.105, which gives
d.
nt
dt
=
_
·
¡olts
2
_ _
T
tr|
+T
J
1
_
= ÷
_
·
¡olts
2J
1
_ __
·
¡olts
2
_
[1
s:
i
s
i
:
] sin(0
nt
) ÷T
tr|
_
.
(2.106)
For a motor T
tr|
is a negative function of speed because it opposes rotor rota-
tion; e.g., with viscous friction T
tr|
= ÷/
}
_
J0me
J|
_
where /
}
is the coe¢cient of
friction. For a generator the applied T
tr|
due to a prime mover may be taken
as a positive constant over a limited range of speed but more precisely it is
a function of speed (
J0me
J|
) given by the prime-mover torque-speed characteris-
tic. In general, a torque-speed characteristic encompasses many e¤ects within a
prime mover including viscous friction associated with ‡uid (e.g., fuel) ‡ow as
well as other mechanical motion. In either case, the functions for T
tr|
can be
represented by a Taylor series if increased accuracy is required.
2.19. SYSTEM ANALYSIS 73
2.19.3 State Variable Concept
The state-variable concept, which is fundamental to system theory, is introduced
brie‡y here as a conceptual aid and as a connection to more advanced topics
which are beyond the scope of this course. For a given system, state variables
are de…ned as any minimal set (vector) of variables with the following properties:
1. the set determines all other system variables;
2. future values of the set follow from the present values of the set, given
present and future inputs.
The …rst property (i.e., dependency of all other variables on state variables)
must hold while minimality implies no direct functional dependencies (i.e., im-
plies independence) amongst the state variables themselves, at an arbitrary
instant in time. The second property indicates that knowledge of past inputs
is not necessary to determine present and future system behavior because the
present values of state variables summarize the past or "history" of the sys-
tem. In addition it can be shown that the energy stored in a system is a direct
function of the state variables.
The "evolution" of state variables (second property) is expressed mathe-
matically by the general (nonlinear and time-varying) state equations in vector
form:
dr
dt
= ) (r(t), n(t), t) (2.107)
and
j (t) = q (r(t), n(t), t) ; (2.108)
where r(t), n(t), and j (t) denote vectors of state variables, inputs, and outputs,
respectively. Equation 2.107 represents a set of …rst-order di¤erential equations
where ) (r(t), n(t), t) can be a nonlinear, time-varying function. Equation 2.108
represents a set of output equations where q (r(t), n(t), t) can be a nonlinear,
time-varying function. An input is a completely constrained or speci…ed vari-
able (function of time) "applied" somewhere in the system; e.g., an ideal voltage
source. An output is a variable which is a function of state variables and in-
put but otherwise completely unconstrained. Input and output roles must be
interpreted or assigned within the context of a given system. Note that state
equations can be simpli…ed by using matrix form if the system is linear but
electromechanical systems generally involve nonlinear terms.
2.19.4 State Variables for Machine Models
The elementary machine model is an idealized lossless magnetic energy storage
system characterized by its rotor and stator spatial current distributions. These
current distributions, hence their magnetic …elds, are: …xed geometrically with
respect to the individual rotor and stator axes; scalable respectively by i
:
and i
s
;
and movable only with respect to each other by rotation based on the parameter
0
nt
. It can be shown that i
:
, i
s
, and 0
nt
are state variables of the elementary
74 CHAPTER 2. ROTATING MACHINES
machine, where 0
nt
is regarded as an arbitrary but …xed parameter as if the
rotor were held in place by a mechanical clamp. This (temporary) constraint on
0
nt
gives
J0me
J|
= 0, necessary due to the assumption that the rotor is massless;
otherwise the magnetic torque T would cause
J0me
J|
÷÷·.
The practical machine model was produced from the elementary machine
model by the addition of external elements: rotor moment of inertia J
1
, a torque
on J
1
, plus winding and load resistances. Only J
1
has any e¤ect with respect
to required state variables, because it is a kinetic-energy storage element rather
than a loss or source element. I.e., now one must introduce the additional state
variable: .
nt
.
=
J0me
J|
. Note that taking J
1
÷÷ · would impose
J.me
J|
= 0
hence
J0me
J|
= .
nt
= 0, in the same vein as the aforementioned mechanical
clamp which is now removed.
The state variables 0
nt
and
J0me
J|
are analogous to the position and velocity of
a mass which undergoes translational motion where a restoring force is provided
by a spring. There potential energy is a function of position and kinetic energy
is a function of velocity. However now the restoring force is provided by the
magnetic …eld which acts like a "spring".
2.19.5 State Equations
State equations for the practical machine in the form of equation 2.107 will
follow from equations 2.82, 2.83, and 2.106. Here the state vector is
r =
_
¸
¸
_
r
1
r
2
r
3
r
4
_
¸
¸
_
where r
1
.
= i
s
, r
2
.
= i
:
, r
3
.
= 0
nt
, and r
4
.
= .
nt
. We require expressions for
Jr1
J|
,
Jr2
J|
,
Jr3
J|
,and
Jr4
J|
as functions of the state variables r
1
, r
2
, r
3
, r
4
, the inputs,
and the machine parameters. Note that the expression for
Jr4
J|
by equation 2.106 and
Jr3
J|
= r
4
. The expressions for
Jr1
J|
.
=
JIs
J|
and
Jr2
J|
.
=
JIr
J|
follow by solving equations 2.82 and 2.83 for the "unknowns"
Jr1
J|
and
Jr2
J|
, where
r
1
, r
2
, r
3
, r
4
, ·
s
, ·
:
and all machine parameters are assumed given or "known";
there
J0me
J|
.
=
Jr3
J|
= r
4
is known as well. These "knowns" (or corresponding
known expressions) play the role of coe¢cients of the "unknowns" or the role
of constants, so the method of solution is analogous to solving a 2-by-2 system
(two equations and two unknowns) of linear equations. The actual solution is
left as an exercise for the reader.
It remains to assign inputs and outputs to the resulting state equations,
based on the intended machine operation as motor or generator. In any case,
one (albeit …xed or slowly varying) input is the applied rotor (…eld) voltage ·
:
because it serves only to establish the machine’s magnetic operating …eld. For
a motor, the other input is the applied stator (armature) voltage ·
s
and the
output is normally taken as the angular position r
3
.
= 0
nt
and/or the angular
velocity r
4
.
= .
nt
, while the external torque T
tr|
is opposing and represents
2.19. SYSTEM ANALYSIS 75
a mechanical load (e.g., viscous friction) as a function of r
3
.
= 0
nt
. For a
generator, the other input is the external applied torque T
tr|
due to a prime
mover and the output is taken as the stator (armature) voltage ·
s
across the
tj
. Hence the output form of equation 2.108 depends on the intended
operation of the machine.
The resulting state equations are nonlinear but amenable to solution by
numerical integration. Note that the coenergy was found already in equation
2.84, as a function of the state variables r
1
.
= i
s
, r
2
.
= i
:
, r
3
.
= 0
nt
.
State equations for the elementary or idealized machine require only the
state variables r
1
.
= i
s
, r
2
.
= i
:
, r
3
.
= 0
nt
and are basically a subset of the state
equations for the practical machine. Here the assumed constraint of r
3
.
= 0
nt
to a constant value is necessary. Otherwise the torque T on the massless rotor
would cause its instantaneous alignment with the stator axis such that r
3
.
=
0
nt
= 0 permanently. This would give zero magnetic torque T and the system
would function thereafter simply as two mutually coupled inductances.
2.19.6 Motor and Generator Operation
It remains to derive a simple criterion which indicates if a machine is operating
as a motor or as a generator at any given time, irrespective of its intended mode
of operation. I.e., motors may be operating as generators (temporarily) and
vice versa. E.g., an AC synchronous generator may be forced into momentary
motor operation to "sync up" when connected to a power grid (so called "in…nite
bus"). E.g., in an electric vehicle regenerative braking with concurrent battery
recharge can be accomplished by temporarily reducing the armature voltage of
a DC motor to zero, in which case the DC motor is acting as a generator.
The desired criterion will follow simply by determining if the magnetic …eld
is delivering or absorbing mechanical power. Instantaneous mechanical power
output (delivered) at the rotor shaft is
j = T.
n
, (2.109)
where T is the instantaneous torque on the rotor produced by the magnetic
…eld; we assume .
n
0, i.e., the machine does not rotate backwards though
it may decelerate occasionally. Clearly power delivery (motor operation) and
power absorption (generator operation) correspond to j 0 (hence T 0) and
j < 0 (hence T < 0), respectively.
The connection between motor operation (T 0) or generator operation
(T < 0) and the machine state is found by interpretation of equation 2.95.
Clearly
_
motor: T 0
generator: T < 0
_
if
_
c
s:
< 0
c
s:
0
_
(2.110)
since 1
s
0 and 1
:
0. Recall that c
s:
is the instantaneous angle of the stator
1-…eld ("‡ux") wave relative to the rotor 1-…eld ("‡ux") wave. Hence the
stator (armature) wave leads or "pulls" the rotor wave in a motor. Conversely
the rotor wave leads or "pulls" the stator (armature) wave in a generator.
76 CHAPTER 2. ROTATING MACHINES
The torque T, hence its sign, is also given by the fundamental equation
2.93 showing the connection to the angle 0
n
. However this equation is more
di¢cult to interpret for the present purpose because 0
n
is the angle between the
rotor and stator magnetic axes which is not known explicitly. A more complete
analysis of 0
n
would be redundant in that it is analogous to the process which
was used to derive equation 2.95.
2.19.7 Application
Complete solution of the system di¤erential equations in their original form or in
state-variable form is di¢cult due to the inherent nonlinear terms. However in
the important AC steady-state case the solution becomes less complex. So many
problems related to speci…c types of machines are considered under AC steady-
state conditions in the sequel. Complete solutions which include transients
require more advanced (usually numerical) techniques.
Chapter 3
Synchronous Machines
3.1 Introduction
A detailed analysis of AC polyphase (actually three-phase) synchronous ma-
chines will be presented here. Some underlying principles of synchronous ma-
chines were given previously in Section 1.11 of the notes entitled Rotating Ma-
chines. In particular, equation 1.18 is the connection between electrical and
mechanical frequencies whereby the rotor rotates in synchronism with the mag-
netic …eld produced by the armature currents.
The analysis here amounts to a generalization of the elementary non-salient
pole machine presented in the aforementioned notes. The polyphase synchro-
nous machine includes three stator (armature) windings (one per phase) and one
rotor winding whereas the elementary machine has one stator winding and one
armature winding. However basic principles carry over intact from the elemen-
tary machine to the polyphase synchronous machine, with regard to winding
self- and mutual inductances, air gap …eld, energy and coenergy, torque, etc.
3.2 Machine Construction
Synchronous machine construction may utilize either a non-salient pole (cylin-
drical) rotor (see Figs. 4.10 and 4.11) or a salient pole rotor (see Fig. 4.9).
Non-salient pole rotors are more suitable for high-speed gas- or steam- turbine
generators. Salient pole rotors are more suitable for slow-speed multi-polar
hydroelectric generators and for most synchronous motors.
3.3 Excitation
DC excitation is applied to the …eld winding which is almost invariably on the
rotor of the synchronous machine. The exciter (machine) is usually mounted
on the same shaft as the rotor. DC exciters (found in older machines) supply
79
80 CHAPTER 3. SYNCHRONOUS MACHINES
the …eld winding through slip rings because their DC output must be taken
from brushes which are stationary with respect to the commutator (on the
rotor of a DC machine). AC exciters (found in newer machines) supply DC by
means of recti…ers where (depending on the con…guration of the exciter) the
supply connection is either through slip rings or rotating with the shaft (hence
a "brushless excitation" system). The excitation power is typically only one to
two percent of the synchronous machine rating.
3.4 Network Operation
The US power grid is a network containing many (e.g., hundreds of) synchro-
nous generators operating in parallel over thousands of miles of transmission
lines. Most electrical energy is produced and distributed this way because of
economic considerations regarding plant investment and operating costs. I.e.,
generating capacity can be shared by di¤erent regions which experience peak
demand at di¤erent times. Likewise reliability should increase because backup
power sources are available if a speci…c machine fails. However a serious network
drawback is possible loss of synchronism due to large transients initiated by a
disturbance; in extreme cases this can lead to a widespread power outage. This
problem is an area of active research in power engineering.
An individual machine operates in a network as follows. The voltage and
frequency at its armature terminals are constrained by the overwhelming pres-
ence of numerous other, synchronized machines over the network. Hence the
individual machine has a component of armature current (magnetic-‡ux wave)
which rotates at the network synchronous frequency ()
t
) and tends to "lock
in" the machine’s mechanical frequency ()
n
). For analytic purposes, the re-
mainder of the network seen by the individual machine can be modeled as a
constant-voltage, constant-frequency source ("in…nite bus").
3.5 Torque
The torque produced by the magnetic …eld at the rotor shaft is
T =
_
¬
2
_
_
·
¡olts
2
_
2

1
1
}
sinc
1J
; (3.1)
where
1
.
= resultant air gap ‡ux (per pole), 1
}
.
= mmf of DC …eld winding,
c
1J
.
= electrical phase angle between magnetic axes of
1
and 1
}
. This equation
follows directly from equation 1.99 in the aforementioned notes simply by the
following identi…cations:
1
!
s:
; 1
}
! 1
:
; and c
1J
! c
:
. The torque
T acts in a direction to bring the magnetic …elds into alignment, as indicated
by the negative sign. (Note that the textbook omits the negative sign with this
understanding.) The torque-angle characteristic of equation 3.1 is plotted in
Fig. 5.1, for
1
and 1
}
constant. In this plot the regions corresponding to
3.6. INDUCTANCE PARAMETERS 81
motor and to generator operation are identi…ed by means of the criterion in
Section 1.19.6 in the aforementioned notes.
The torque angle c
1J
must increase if the machine torque T must increase,
in view of equation 3.1. This occurs for a motor if the opposing torque increases
due to a greater mechanical load or for a generator if the applied prime mover
torque must increase to supply a greater electrical load. A sudden (e.g., step)
increase in mechanical or electrical load will cause a "hunting transient"; i.e.,
a damped mechanical oscillation of the rotor about its new steady-state torque
angle. Analysis of the hunting transient of an individual machine as it pulls
into synchronism immediately after connection to an in…nite bus will be treated
later.
Clearly the maximum jTj available occurs for sinc
1J
= 1 hence c
1J
=
90

. Synchronism will be lost if the machine torque needs to exceed this value
to retain synchronism against the load. I.e., then the machine "pulls out" of
synchronism, so maximum jTj is called the pull-out torque. In a generator loss
of synchronism (pull-out) can result in dangerous overspeed operation so sensors
are usually applied to shut down the machine.
3.6 Inductance Parameters
The ‡ux linkages of the two-pole three-phase machine in Fig. 5.2 are given in
matrix form by:
_
¸
¸
_
`
o
`
b
`
c
`
}
_
¸
¸
_
=
_
¸
¸
_
L
oo
L
ob
L
oc
L
o}
L
bo
L
bb
L
bc
L
b}
L
co
L
cb
L
cc
L
c}
L
}o
L
}b
L
bc
L
}}
_
¸
¸
_
_
¸
¸
_
i
o
i
b
i
c
i
}
_
¸
¸
_
(3.2)
where the subscripts a, /, c indicate armature (stator) phase variables and the
subscript ) indicates …eld (rotor) variables. The script L denotes the possibility
that an inductance may vary with rotor position 0
nt
, while like subscripts (as
in L
oo
) indicate a self-inductance and unlike subscripts (as in L
ob
) indicate a
mutual inductance. Mutual inductance between windings will vary with 0
nt
if
one subscript is from the set fa, /, cg and the other subscript is from the set f)g:
e.g., L
o}
, L
b}
, L
}o
, etc. In this case the windings lie on opposite sides (stator
side vs. rotor side) of the gap. Self- or mutual inductance of windings will not
vary with 0
nt
(in this uniform-gap machine) if both subscripts are from the set
fa, /, cg or if both subscripts are from the (trivial) set f)g: e.g., L
oo
, L
ob
, L
}}
,
etc. In this case both windings lie on the same side of the gap. In the sequel,
inductances (or parameters) which are not a function of 0
nt
will be denoted by
an italic 1 rather than a script L.
Note that equation 3.2 is analogous to equation 1.74 which gives the ‡ux
linkages for the elementary machine in the aforementioned notes. Now the
subscript : corresponding to the single stator winding of the elementary machine
has been generalized to the subscripts fa, /, cg which correspond to three stator
phase windings. Also the subscript r corresponding to the single rotor winding
82 CHAPTER 3. SYNCHRONOUS MACHINES
of the elementary machine has been changed to ) to denote a single rotor …eld
winding.
3.7 Self-Inductances
The stator self-inductances L
oo
, L
bb
, L
cc
and the rotor self-inductance L
}}
are
not functions of 0
nt
, in view of the above discussion. Each of these self-
inductances can be decomposed into two parts: one due to the spatial fun-
damental component of air-gap ‡ux; and the other due to leakage. Note that
harmonic e¤ects due to stator or rotor slot openings are neglected here.
The corresponding decompositions are:
L
oo
= 1
oo
= 1
oo0
+1
ol
; (3.3)
L
bb
= 1
bb
= 1
bb0
+1
bl
; (3.4)
L
cc
= 1
cc
= 1
cc0
+1
cl
; (3.5)
and
L
}}
= 1
}}
= 1
}0
+1
}l
. (3.6)
Here the appended subscripts
0
and
l
denote the parts due to the spatial fun-
damental component and leakage, respectively. The calculation of these induc-
tances due to spatial fundamental components is given in Appendix B of the
textbook.
3.8 Mutual Inductances
The mutual inductance between windings on the same side of the gap need be
considered only for the stator because the rotor has only one winding. These
mutual inductances are:
L
bo
= L
ob
= 1
ob
= [1
oo0
cos (c)]
o=+120
=
_
1
2
_
1
oo0
; (3.7)
L
co
= L
oc
= 1
oc
= [1
oo0
cos (c)]
o=+120
=
_
1
2
_
1
oo0
; (3.8)
and
L
bc
= L
cb
= 1
cb
= [1
oo0
cos (c)]
o=+120
=
_
1
2
_
1
oo0
. (3.9)
These results for mutual inductance are derived in Appendix B of the text-
book. Note that a derivation of only one result (e.g., L
ob
) su¢ces because the
remaining results follow by symmetry arguments. Also note the simple rela-
tion between mutual inductance and self-inductance (e.g., 1
ob
and 1
oo0
) which
occurs because all windings are on the same (stator) core. I.e., the mutual in-
ductance derives from the fraction (including sense or "polarity") of the total
(self-inductive) ‡ux of one phase winding which links to another phase winding.
3.9. SYNCHRONOUS INDUCTANCE 83
The mutual inductance between windings on opposite sides of the gap are:
L
}o
= L
o}
= 1
o}
cos (0
nt
) ; (3.10)
L
}b
= L
b}
= 1
o}
cos (0
nt
120

) ; (3.11)
and
L
}c
= L
c}
= 1
o}
cos (0
nt
+ 120

) . (3.12)
These results for mutual inductance including calculation of 1
o}
are derived in
Appendix B of the textbook. Again a derivation of only one result (e.g., L
o}
)
su¢ces because the remaining results follow by symmetry arguments. These
mutual inductances are a function of 0
nt
which is the necessary mechanism for
electromechanical energy conversion to occur.
3.9 Synchronous Inductance
The operation of each phase can be analyzed separately (i.e., on a per-phase
basis) by means of the so-called synchronous inductance which results under
conditions of balanced (normally steady-state) three-phase armature currents.
In other words, mathematically the synchronous inductance incorporates the
mutual (inductive) interaction between phase windings into a single per-phase
model. It su¢ces to derive this model for phase a because identical phase / and
c models follow directly by symmetry arguments.
We begin by writing out the phase-a ‡ux expression:
`
o
= L
oo
i
o
+L
ob
i
b
+L
oc
i
c
+L
o}
i
}
(3.13)
which follows from the …rst row of the matrix equation 3.2. This yields
`
o
= (1
oo0
+1
ol
) i
o
+
_

_
1
2
_
1
oo0
_
(i
b
+i
c
) +L
o}
i
}
(3.14)
by substitution of the previously derived expressions for inductances from equa-
tions 3.3, 3.7, and 3.8. The assumption (normally given in phasor form) that
the three-phase armature currents are balanced is equivalent to an instantaneous
balance in the time domain:
i
o
+i
b
+i
c
= 0 !i
o
= (i
b
+i
c
) . (3.15)
Hence equation 3.14 reduces to
`
o
= 1
s
i
o
+L
o}
i
}
(3.16)
by substitution of equation 3.15 for i
o
, where we identify the constant synchro-
nous inductance
1
s
.
=
_
3
2
_
1
oo0
+1
ol
. (3.17)
Here the factor
_
3
2
_
corresponds to the same factor in the forward stator mmf
wave which results under balanced conditions.
84 CHAPTER 3. SYNCHRONOUS MACHINES
3.10 Armature Equivalent Circuit
The per-phase (line-to-neutral) equivalent circuit at the phase-a armature ter-
·
o
= 1
o
i
o
+c
o
(3.18)
where the series DC armature resistance 1
o
is now included and the induced
voltage is c
o
.
=
JXa
J|
. From equation 3.16
c
o
.
=
d`
o
dt
=
d
dt
(1
s
i
o
+L
o}
i
}
) =
d
dt
(1
s
i
o
) +
d
dt
(L
o}
i
}
) = c
oo
+c
o}
(3.19)
where
c
oo
.
=
d
dt
(1
s
i
o
) = 1
s
di
o
dt
(3.20)
and
c
o}
.
=
d
dt
(L
o}
i
}
) = L
o}
di
}
dt
+
_
dL
o}
dt
_
i
}
. (3.21)
By equation 3.10
L
o}
= 1
o}
cos (.
t
t +c
t0
) (3.22)
where
0
nt
=
_
_
·
¡olts
2
_
Þ
poles
=2
_
0
n
= .
t
t +c
t0
, (3.23)
so
dL
o}
dt
= .
t
1
o}
sin(.
t
t +c
t0
) . (3.24)
This gives
c
o}
= .
t
1
o}
1
}
sin(.
t
t +c
t0
) (3.25)
by substitution into equation 3.21 with DC …eld excitation current i
}
= 1
}
hence
JI
f
J|
= 0. Now by equation 3.19 with substitution of equations 3.20 and 3.25 we
have
c
o
= 1
s
di
o
dt
+c
o}
(3.26)
and …nally
·
o
= 1
o
i
o
+1
s
di
o
dt
+c
o}
(3.27)
by substitution of the previous equation into equation 3.18. The per-phase
equivalent circuit at the phasea armature terminals, corresponding to equation
3.27 and shown in Fig. 5.3, consists of three series elements: the resistance 1
o
,
the synchronous inductance 1
s
, and the generated (speed) voltage c
o}
. Figs.
5.3a and 5.3b show reference current directions for i
o
as inward and outward
for conventional motor and generator operation, respectively.
For AC steady-state operation the relation between voltage and current at
the phasea armature terminals is now restated in RMS-phasor form. I.e.,
conversion of equation 3.27 to phasor form yields
´
\
o
= 1
o
´
1
o
+,A
s
´
1
o
+
´
1
o}
(3.28)
3.10. ARMATURE EQUIVALENT CIRCUIT 85
where
´
\
o
,
´
1
o
,
´
1
o}
and \
o
, 1
o
, 1
o}
denote phasors and their RMS magnitudes,
respectively, and the synchronous reactance is
A
s
.
= .
t
1
s
. (3.29)
(Note that A
s
is indicated for 1
s
in Fig. 5.3.) The phasea armature current
is
i
o
=
__
p
2
_
1
o
_
cos (.
t
t) (3.30)
so
´
1
o
= 1
o
c
¸0
(3.31)
is the phase reference. By equation 3.25 for the generated voltage c
o}
, where
max fc
o}
g = .
t
1
o}
1
}
, we have
1
o}
=
.
t
1
o}
1
}
p
2
(3.32)
and hence the phasor representation
´
1
o}
= ,1
o}
c
¸oe0
. (3.33)
The synchronous reactance decomposes, by substitution of equation 3.17
into equation 3.29, into
A
s
= A
,
+A
ol
, (3.34)
where
A
,
.
= .
t
_
3
2
_
1
oo0
(3.35)
is the e¤ective magnetizing reactance and
A
ol
.
= .
t
1
ol
, (3.36)
is the leakage reactance. Here A
,
corresponds to the forward fundamental
spatial ‡ux wave produced by the three armature phase currents under balanced
conditions. The alternative form of the equivalent circuit, which corresponds to
this decomposition, is given in Fig. 5.4. The voltage
´
1
1
.
= ,A
,
´
1
o
+
´
1
o}
(3.37)
generated by the total air gap ‡ux is called the air-gap voltage or voltage behind
leakage reactance; it is labelled in Fig. 5.4.
Refer to Examples 5.1 and 5.2. The typical order of magnitude of the im-
pedance components (on a per-unit basis) is discussed on page 256 of the text-
book.
86 CHAPTER 3. SYNCHRONOUS MACHINES
3.11 Open-Circuit Characteristic
The open-circuit characteristic (OCC) is \
o
as a function of 1
}
, under the con-
dition 1
o
= 0 (open circuit or no electrical load). Under this condition \
o
= 1
o}
by equation 3.28 so
\
o
=
.
t
1
o}
1
}
p
2
(3.38)
by equation 3.32. Hence
1
o}
=
p
2\
o
.
t
1
}
(3.39)
can be determined by measurement.
The OCC is analogous to a magnetization curve for a DC machine because
core saturation e¤ects can introduce a downward bend in the characteristic. It
relates the spatial fundamental of the air-gap ‡ux to the mmf over the magnetic
path, due to 1
}
acting alone. The OCC starts linearly at the origin where
saturation has not yet occurred and air-gap reluctance dominates the inductance
1
o}
. The linear extrapolation or extension of the OCC from the origin is called
the air-gap line, which is normally used for comparative purposes.
Refer to Fig. 5.5 for an illustration of the OCC including an air-gap line.
An OCC may use either absolute or per-unit values. In the latter case the base
voltage is usually taken as the rated voltage of the machine. Refer to Example
5.3.
Machine losses under open-circuit conditions (no electrical load) consist of
the mechanical power necessary to overcome friction and windage at synchro-
nous rotational speed plus core losses due to magnetic ‡ux. Hence core losses
can be determined by the di¤erence between machine loss measured with and
without …eld excitation. Refer to Fig. 5.6 for a typical curve of core loss vs. \
o
;
here \
o
is varied by changing 1
}
.
3.12 Short-Circuit Characteristic
The short-circuit characteristic (SCC) is 1
o
as a function of 1
}
, under the con-
dition \
o
= 0 (short-circuit armature terminals). Under this condition
´
1
o}
= 1
o
´
1
o
+,A
s
´
1
o
= (1
o
+,A
s
)
´
1
o
(3.40)
by equation 3.28 using the generator reference direction for current
´
1
o
, so
(1
o
+,A
s
) =
´
1
o}
´
1
o
. (3.41)
Usually 1
o
A
s
in which case the previous equation gives
A
s
'
1
o}
1
o
; (3.42)
3.13. MEASUREMENT OF SYNCHRONOUS REACTANCE 87
so A
s
can be determined from 1
o}
on the OCC and 1
o
on the SCC, where the
…eld current 1
}
must be the same for both.
SCC is linear if the machine is unsaturated. Noting that equation 3.40
expresses KVL around the loop where \
o
= 0 (instead of the machine rating),
the air-gap voltage
´
1
1
in equation 3.37 and hence the air-gap ‡ux will be a small
fraction (e.g., 0.15) of its normal value. In other words the armature reaction
‡ux wave is moving with and opposing the ‡ux wave due to the rotating …eld
winding. This implies that the machine is operating in an unsaturated condition.
The linear SCC is derived as follows. First we have
,1
o}
c
¸oe0
= (1
o
+,A
s
)
´
1
o
(3.43)
by equations 3.33 and 3.40 so
,
_
.
t
1
o}
1
}
p
2
_
c
¸oe0
= (1
o
+,A
s
)
´
1
o
(3.44)
by substitution of equation 3.32 for 1
o}
. Solution of this last equation for
´
1
o
gives
´
1
o
=
_
.
t
1
o}
1
}
p
2 (,1
o
+A
s
)
_
c
¸oe0
, (3.45)
hence the linear relation
1
o
=

¸
¸
¸
.
t
1
o}
p
2 (,1
o
+A
s
)
¸
¸
¸
¸
_
1
}
(3.46)
or
1
o
=
_
1
p
2
_ _
.
t
1
o}
A
s
_
1
}
(3.47)
if 1
o
A
s
. Refer to Fig. 5.9 which illustrates OCC and SCC.
Machine losses under short-circuit conditions consist of the mechanical power
necessary to overcome friction and windage (determined from an open-circuit
test with 1
}
= 0) at synchronous rotational speed plus losses due to armature
current. Core loss is considered negligible because the magnetic ‡ux level is low
under short-circuit conditions. Hence the 1
2
1 losses due to armature current
can be found as the di¤erence between machine losses under short-circuit and
open-circuit conditions. Refer to Fig. 5.10 for a plot of short-circuit load loss.
Refer to Example 5.5. Refer to pp. 263-65 of textbook for a complete discussion
of these losses.
3.13 Measurement of Synchronous Reactance
First, SCC is determined under short-circuit conditions created by application
of a three-phase short-circuit with appropriate current sensors at the armature
terminals of the machine. The …eld current is adjusted to obtain a range of
88 CHAPTER 3. SYNCHRONOUS MACHINES
armature current 1
o
not to exceed the machine rating. OCC is determined
under open-circuit conditions by measurement of 1
o}
vs. 1
}
.
The unsaturated synchronous reactance A
s
follows from equation 3.42 by
substitution of the values for 1
o
and 1
o}
read from the SCC and the OCC
at the same value of 1
}
. 1
o}
must be read from the air-gap line on the OCC
because the SCC is under unsaturated conditions. This process is analogous to
determination of the series impedance in a Thevenin equivalent circuit, which
is not surprising as the armature equivalent circuit is already in Thevenin form.
Refer to Fig. 5.9 which illustrates OCC and SCC for the determination of A
s
.
Refer to pp. 261-62 and Fig. 5.9 for a discussion of saturated synchronous
reactance. Refer to Example 5.4.
3.14 Short-Circuit Ratio
The short-circuit ratio (SCR) is
oC1
.
=
"AFNL"
¸ .. ¸
[i
}
]
Occ:\a at rated value
[i
}
]
Scc:1a at rated value
. ¸¸ .
"AFSC"
; (3.48)
where "AFNL" and "AFSC" mean "amperes …eld no load" and "amperes …eld
short circuit", respectively; we will show
oC1 =
1
A
s, per unit
. (3.49)
Substitution for i
}
in equation 3.48, with i
}
determined under the numerator
and denominator conditions by equations 3.38 and 3.47 respectively, gives
oC1 =
(
p
2)\a
.eJ
af
_
(
p
2)1a
.eJ
af
_
A
s
=
1
·s
(
Va
Ia
)
=
1
A
s, per unit
.
Here \
o
and 1
o
were taken as rated values for the machine so
_
\a
1a
_
is the
impedance base for conversion to per unit impedance. Refer to pp. 261-62 of
textbook for interpretation of A
s
under saturated conditions in machine.
3.15 Field Winding Flux
The …eld winding ‡ux linkage is constant under DC …eld excitation assuming
steady-state (synchronous) machine operating conditions with balanced three-
phase armature currents. I.e., the armature ‡ux wave has only a forward com-
ponent which rotates in synchronism with the rotor given balanced three-phase
armature currents. Hence the …eld ‡ux linkage due to the armature currents is
constant (non time-varying) and the armature currents do not induce a voltage
in the …eld winding. This result is equivalent to that which would be obtained
by writing out and evaluating the …eld ‡ux expression for `
}
from equation 3.2.
A voltage can be induced into the …eld winding under transient unbalanced
conditions on the armature currents. Machine performance can be a¤ected
under these conditions. Also, an initial (11) transient occurs upon application
of the DC …eld excitation, such that the …eld current settles to its …nal DC value
1
}
=
\
f
1
f
.
We determine the maximum power that can be delivered by a synchronous
machine without loss of synchronism due to excess torque requirements. Here
a synchronous machine with known armature (Thevenin) equivalent circuit is
connected to an external system represented by another Thevenin equivalent cir-
cuit, as shown in Fig. 5.12. The external system may contain more synchronous
machines, transmission lines, and transformer banks.
This problem is a special case of power ‡ow between two voltage sources
´
1
1
and
´
1
2
through a series impedance
7 = 1 +,A = j7j c
¸ç
Z
!c
2
.
= arctan
_
A
1
_
, (3.50)
as shown in the network of Fig. 5.11a. Here 7 represents the series combination
of Thevenin equivalent impedances of the machine and the external system. In
phasor terms, the "receiving" voltage is
´
1
2
.
= 1
2
c
¸0
= 1
2
, (3.51)
which serves as the phase reference, the "sending" voltage is
´
1
1
.
= 1
1
c
¸o
, (3.52)
and the series-element current is
´
1
.
= 1
2
c
¸ç
=
´
1
1

´
1
2
7
(3.53)
with phase angle c determined by the network. The real power delivered to the
receiving end is
1
2
= 1
2
1 cos (c) (3.54)
so we will …nd
´
1 to determine 1 cos (c) . By substitution of equations 3.52, 3.51,
and 3.50 into equation 3.53:
´
1 =
1
1
c
¸o
1
2
j7j c
¸ç
Z
=
1
1
c
¸(oç
Z
)
j7j

1
2
c
¸ç
Z
j7j
(3.55)
90 CHAPTER 3. SYNCHRONOUS MACHINES
which gives
Re
_
´
1
_
= 1 cos c =
1
1
cos (c c
2
)
j7j

1
2
cos (c
2
)
j7j
. (3.56)
This yields
1 cos c =
1
1
cos (c c
2
)
j7j

1
2
1
j7j
2
(3.57)
because
cos (c
2
) = cos (c
2
) =
1
j7j
. (3.58)
Finally
1
2
=
1
1
1
2
cos (c c
2
)
j7j

1
2
2
1
j7j
2
, (3.59)
by equation 3.54 with substitution of equation 3.57, or alternatively
1
2
=
1
1
1
2
sin(c +c
2
)
j7j

1
2
2
1
j7j
2
(3.60)
where
c
2
.
= 90

c
2
= arctan
_
1
A
_
. (3.61)
Likewise the power delivered by the sending end is
1
1
=
1
1
1
2
sin(c c
2
)
j7j
+
1
2
1
1
j7j
2
. (3.62)
It is important to note that
1
1
' 1
2
'
1
1
1
2
sin(c)
A
(3.63)
if 1 j7j because that implies j7j ' A and c
2
' 0. Equation 3.63 is called
the power angle characteristic and c is called the power angle for synchronous
machines.
Maximum power transfer
1
1
' 1
2
'
1
1
1
2
A
(3.64)
occurs at c = 90

if 1
1
and 1
2
are constant. Further 1
1
2
if
c 0 (c < 0) and power ‡ow is from 1
1
to 1
2
(1
2
to 1
1
).
For the synchronous machine shown in Fig. 12, equation 3.63 yields
1 =
1
o}
\
JQ
sin(c)
A
S
+A
JQ
(3.65)
by the identi…cations 1
o}
!1
1
, \
JQ
!1
2
, and (A
S
+A
JQ
) !A. Recall
that this is a per-phase analysis so total power is 1
T
= 31. It is possible to apply
equation 3.63 to di¤erent subsections of the network by using the alternative
identi…cation 1
o}
!1
1
, \
o
!1
2
, A
S
!A or the alternative identi…cation
\
o
! 1
1
, \
JQ
! 1
2
, and A
JQ
! A. However \
o
! 1
1
is not constant
as 1 changes, so an additional relation would be required to relate \
o
to 1.
The power-angle characteristic in equation 3.65 and the torque in equation
3.1 have the same form with respect to the power angle c. This con…rms that
the limiting factor in power transfer is the maximum available torque. Equation
3.65 indicates that power transfer can be increased by increasing 1
o}
if \
JQ
is
a …xed system-bus voltage. However 1
o}
is ultimately limited due to thermal
factors regarding the …eld current 1
}
. Also generators are usually operated at a
power angle c signi…cantly less than 90

Refer to Examples 5.6 and 5.7.
Steady state operating characteristics refer to relations between \
o
, 1
}
, 1
o
, power
factor, power, and e¢ciency.
Compounding curves relate 1
}
to 1
o
under the constraint that \
o
is constant
as load power 1 varies, with various …xed power factors as a parameter. Refer to
p. 275 of textbook for additional details and Fig. 5.15 which illustrates typical
compounding curves.
Capability curves give the locus of maximum real power 1 and reactive power
Q for a machine operating at rated voltage \
o
, due to simultaneous limits on 1
o
and 1
}
imposed by heating of armature and …eld windings respectively. Refer
to pp. 276-78 of the textbook for a discussion and Figs. 5.16 and 5.17 which
illustrate a typical capability curve and its construction, respectively.
A brief derivation or construction of the capability curve follows with ref-
erence to Fig. 5.17. First assume the limiting value 1
o
(a magnitude without
constraint on the phase of
´
1
o
) and operation at rated \
o
. Recall
o = 1 +,Q = \
o
1
o
(3.66)
which implies
joj = j\
o
j j1
o
j = constant. (3.67)
Hence
joj
2
= 1
2
+Q
2
= constant (3.68)
so this locus is a circle in the 1 Q plane with center at the origin and radius
joj KVA. Next assume the limiting value 1
}
which implies a certain limiting
value 1
o}
which we will use instead of 1
}
; any restriction on 1
o
is removed now.
We have
´
\
o
= ,A
s
´
1
o
+
´
1
o}
(3.69)
92 CHAPTER 3. SYNCHRONOUS MACHINES
from equation 3.28 using the generator reference direction for
´
1
o
and assuming
1
o
= 0; hence
´
1
o
=
´
\
o

´
1
o}
,A
s
=
,
_
´
\
o

´
1
o}
_
A
s
. (3.70)
We assume
´
\
o
= \
o
is the phase reference here and compute:
1 ,Q =
´
\
o
´
1
o
=
´
\
o
_
_
,
_
´
\
o

´
1
o}
_
A
s
_
_
=
,\
2
o
A
s

,\
o
´
1
o}
A
s
(3.71)
which gives
1 ,
_
Q+
\
2
o
A
s
_
=
,\
o
´
1
o}
A
s
(3.72)
by simple algebraic rearrangement. Note that
´
1
o}
= (1
o}
) c
¸o
(3.73)
where the magnitude 1
o}
is …xed or limited but the angle c is arbitrary. However
¸
¸
¸
¸
¸

,\
o
´
1
o}
A
s
¸
¸
¸
¸
¸
=
j\
o
j
¸
¸
¸
´
1
o}
¸
¸
¸
jA
s
j
=
\
o
1
o}
A
s
(3.74)
so taking the square of the magnitude of both sides of equation 3.72 yields
1
2
+
_
Q+
\
2
o
A
s
_
2
=
_
\
o
1
o}
A
s
_
2
.
This locus is a circle with center at
_
0,
\
2
a
·s
_
in the 1 Q plane and radius
\aJ
af
·s
. The valid operating region is the area in the 1 Q plane where both
constraints are satis…ed simultaneously. This region is the intersection of the
areas enclosed by the two circular loci with the added requirement 1 0, as
shown in Fig. 5.17. Note that the machine rating (apparent power and power
factor) is commonly speci…ed as the intersection of the two circles.
V curves relate 1
}
to 1
o
under the constraint that \
o
constant with various …xed power factors as a parameter. Refer to pp. 278-79 of
textbook for additional details and Fig. 5.18 which illustrates typical V curves.
Chapter 4
Polyphase Induction
Machines
4.1 Introduction
A detailed analysis of AC polyphase (three-phase) non-salient pole induction
machines will be presented here. Some underlying principles of induction ma-
chines were given previously in Section 1.11.3 of the notes entitled Rotating
Machines.
Many principles for analysis of an induction machine carry over from analysis
of a synchronous machine. The stator (armature) windings of three-phase in-
duction machines and three-phase synchronous machines are very similar if not
identical. In either case there is only a forward-rotating armature ‡ux-density
wave if stator currents are balanced. However the rotor ("…eld") winding(s) of
an induction machine and a synchronous machine are fundamentally di¤erent by
design: the former are loops closed by either a short circuit or external resistance
(hence no source applied); whereas the latter are driven by an external source to
establish the operating …eld. In an induction machine the rotor currents (hence
"operating …eld") are generated by induction (generalized transformer action)
from the armature ‡ux-density wave due to slip (di¤erence in rotational speed of
the ‡ux-density wave and rotor winding). This is the fundamental mechanism
by which electromechanical energy conversion occurs.
4.2 Rotor Construction
Wound rotors and squirrel-cage rotors are the two possibilities in induction
machines. A wound rotor is wound with the same number of poles as the stator
whereas a squirrel-cage rotor will automatically acquire that number of poles
by induction from the stator ‡ux-density wave. The number of rotor phases
is often but not always the same as the number of stator phases. A wound
93
94 CHAPTER 4. POLYPHASE INDUCTION MACHINES
rotor allows connection of external series resistance through slip rings whereas a
squirrel-cage rotor does not because it is permanently, internally short-circuited.
The advantage of a wound rotor is that external resistance can be used to
control certain motor characteristics such as torque versus speed. Its major
disadvantage is cost and reduced reliability due to the presence of slip rings.
Hence squirrel-cage rotors are by far the most common. In some squirrel-cage
rotor designs electromagnetic …eld phenomena are used to vary the internal rotor
resistance automatically with speed, to produce a more favorable torque-speed
characteristic.
Refer to Figs. 6.1, 6.2, and 6.3 which show machines designed with wound
and squirrel-cage rotors.
4.3 Slip
The fractional slip is
:
.
=
:
s
:
:
s
(4.1)
where :
s
is the synchronous rotational rate (RPM) of the armature ‡ux-density
wave and : is the mechanical rotational rate (RPM) of the rotor. This equation
gives
: = (1 :) :
s
; (4.2)
from which it follows that
.
m
= (1 :) .
s
(4.3)
where .
s
and .
m
are the synchronous (magnetic ‡ux-density wave) and me-
chanical (rotor) angular frequencies (where one revolution corresponds to 2¬
radians), respectively. Denoting the slip frequency of the rotor by .
r
.
= :.
s
,
the last equation yields
.
r
= .
s
.
m
= :.
s
(4.4)
or
)
r
.
=
.
r

= :)
e
(4.5)
where )
e
= )
s
.
=
!s
2
and )
m
.
=
!m
2
. (The notation in equation 1.20 of the
aforementioned notes di¤ers from that used here.)
4.4 Rotor Induction
Rotor voltages induced by the rotating stator ‡ux-density wave (Faraday’s law)
will have electrical frequency )
r
(which corresponds to .
r
), because that is the
frequency at which the stator ‡ux appears to vary when viewed from the rotor
(due to relative motion). The rotor currents created by these induced voltages
will cause a rotor mmf wave, hence rotor ‡ux-density wave, which advances (in
the same direction as the stator ‡ux-density wave) around the rotor at frequency
)
r
. Note that the rotor ‡ux-density wave rotates at )
s
when viewed from the
4.5. GENERALIZED TRANSFORMER 95
stator because )
r
+ )
m
= )
s
by equation 4.4 irrespective of any particular
value of slip :. Hence the rotor and stator …elds (‡ux-density waves) rotate in
synchronism.
4.5 Generalized Transformer
Transformer analysis is generalized here to admit relative motion between the
secondary (rotor) and primary (stator) windings. I.e., the basic transformer
equivalent circuit is applied with some external modi…cations in the secondary
loop to account for slip (relative motion). We consider the forward stator ‡ux-
density wave due to each phase separately (per-phase analysis), where all phases
operating simultaneously under balanced conditions will produce a net forward
‡ux-density wave (and no backward ‡ux-density wave).
Consider the stator and rotor ‡ux linkages (for one phase):
`
1
= 1
11
i
1
. ¸¸ .
N111
+1
12
i
0
2
. ¸¸ .
N112
(4.6)
and
`
0
2
= 1
21
i
1
. ¸¸ .
N221
+1
22
i
0
2
. ¸¸ .
N222
; (4.7)
where unprimed and primed variables indicate stator and rotor quantities, re-
spectively, and the turns ratio is
:
.
=
·
1
·
2
. (4.8)
The mutual ‡ux
21
, where
12
=
21
, corresponds to the rotating sinusoidal
stator ‡ux-density wave in the gap,
1l
.
=
11

21
and
2l
.
=
22

21
are
the the stator and rotor leakage ‡uxes, and
c
1
=
d`
1
dt
=
_
·
1
d
11
dt

1
d
12
dt
_
(4.9)
c
0
2
=
d`
2
dt
=
_
·
2
d
21
dt

2
d
22
dt
_
, (4.10)
give the stator and rotor voltages, respectively. This situation is modeled by
the standard transformer equivalent circuit in Fig. 1, where the extracted : : 1
ideal transformer represents the e¤ects of the mutual ‡ux
21
which imposes
the constraints:
´
1
00
2
=
_
1
:
_
´
1
2
(4.11)
and
´
1
0
2
= :
´
1
2
. (4.12)
96 CHAPTER 4. POLYPHASE INDUCTION MACHINES
The e¤ect of slip between rotor and stator is to modify the time derivative of

21
which appears in equation 4.10:
d
21
dt
!:
_
d
21
dt
_
(4.13)
so
´
1
0
2
= :
´
1
00
2
(4.14)
r
= :)
e
or .
r
= :.
e
is the frequency of the induced
rotor voltage
´
1
0
2
. The e¤ect in equation 4.14 is represented by the multiplier :
inserted at the secondary of the ideal transformer in Fig. 1.
The impedance
7
0
2
.
= 1
0
2
+,A
0
2
(4.15)
on the rotor side represents the extracted rotor resistance 1
0
2
and rotor reactance
A
0
2
= .
r
1
0
2l
= :.
e
1
0
2l
(4.16)
due to the rotor leakage inductance 1
0
2l
, where 1
0
2
and 1
0
2l
are physical constants
due to the rotor construction. Note that A
0
2
is proportional to slip : due to the
change in the frequency of the induced voltage
´
1
0
2
.
The equivalent impedance seen on the stator (primary) side of the ideal
transformer is
7
2
.
=
´
1
2
´
1
2
(4.17)
where
´
1
2
=
_
:
:
_
´
1
0
2
, (4.18)
by equations 4.11 and 4.14, and
´
1
2
=
´
1
0
2
:
(4.19)
by equation 4.12. So
7
2
=
_
n
s
_
´
1
0
2
b
I
0
2
n
=
_
:
2
:
_
´
1
0
2
´
1
0
2
=
_
:
2
:
_
7
0
2
(4.20)
by substitution of the previous two expressions into equation 4.17, where
7
0
2
.
= 1
0
2
+,:.
e
1
0
2l
(4.21)
by equations 4.15 and 4.16. Substitution of this last expression into the previous
equation yields
7
2
=
_
:
2
:
_
(1
0
2
+,:.
e
1
0
2l
) =
:
2
1
0
2
:
+,:
2
.
e
1
0
2l
=
1
2
:
+,A
2
(4.22)
4.6. EQUIVALENT CIRCUIT 97
where
1
2
.
= :
2
1
0
2
(4.23)
and
A
2
.
= :
2
.
e
1
0
2l
. (4.24)
Here 1
2
and A
2
are the resistance and reactance seen in 7
2
(stator side) when
the rotor is stationary (: = 1). Equation 4.22 for 7
2
indicates that the reactance
remains constant as A
2
while only the resistance
R2
s
changes, as slip : varies.
The former is due to o¤setting e¤ects:
´
1
0
2
decreases with slip while A
0
2
also
decreases with slip.
The preceding analysis can be applied to either wound rotors or squirrel-
cage rotors with appropriate interpretation. For a wound rotor, where turns are
inherently in series, equations 4.6 and 4.7 can be interpreted by using equivalent
turns ·
1
and ·
2
which represent the net e¤ect of the current sheet induced
across the winding distribution by the rotating ‡ux-density wave . On the other
hand a squirrel-cage rotor is a di¤erent con…guration which may be viewed as
a collection of single conductor loops distributed at uniform angular increments
around the rotor. These loops happen to be joined by the end plates which
renders their operation similar to a parallel rather than a series connection. One
loop can be analyzed directly starting from equations 4.6 and 4.7. The net e¤ect
of all these individual loop currents induced by the rotating ‡ux-density wave
is to approximate a sinusoidal current sheet. The equivalent stator (primary)
current follows by integration over the current sheet so the general form of 7
2
will be the same as before.
4.6 Equivalent Circuit
The per-phase equivalent circuit in a …nal form follows by inserting 7
2
from
equation 4.22 for the equivalent impedance seen from the stator side in Fig. 1.
This is shown in Fig. 6.9 which is a development from Figs. 6.7 and 6.8. Note
that resistances 1
1
and 1
c
have been included to represent primary winding
resistance and core losses, respectively. This equivalent circuit can be applied to
determine many characteristic relations in induction machines between current,
speed, losses, torque, etc.
4.7 Current and Flux
First we review the process by which the stator ‡ux-density wave is produced.
Three stator phase windings spatially displaced by 120

are driven by balanced
three-phase currents. Each phase winding is individually distributed to approx-
imate a sinusoidal current sheet and the net e¤ect is to produce a single forward
rotating sinusoidal current sheet. The application of Ampere’s law to determine
mmf amounts to integration of the sinusoidal current sheet, so the correspond-
ing mmf wave lags the rotating current sheet by 90

. This mmf wave drives
98 CHAPTER 4. POLYPHASE INDUCTION MACHINES
the sinusoidal mutual stator ‡ux-density wave in the gap without any further
phase shift, to produce
d21
dt
and most of
d11
dt
in equations 4.10 and 4.9. The
rotor rotating voltage wave due to the balanced set of three induced rotor phase
voltages c
0
2
is in phase with the ‡ux-density wave but with opposite polarity, by
equation 4.10.
Now consider the phase of the rotating current sheet produced in the rotor
by the induced rotating voltage wave. It is su¢cient to consider only one of the
induced phase voltages c
0
2
to determine the phase c by which the rotating voltage
wave leads the rotating current sheet. The phase c will be the phase between the
individual rotor voltage
´
1
0
2
and the individual rotor current
´
1
0
2
or equivalently
the phase between
´
1
2
and
´
1
2
. (The latter statement follows because equations
4.18 and 4.19 indicate that these primed and unprimed voltage-current pairs are
related by simple scaling.) The phase c between
´
1
2
and
´
1
2
is simply the angle
of the impedance 7
2
:
c = \7
2
, (4.25)
by equation 4.17; equivalently we can write 7
2
in polar form:
7
2
= j7
2
j c
j
. (4.26)
The phase c follows from equation 4.22 or from the equivalent circuit in Fig.
6.9:
c = arctan
_
A
2
R2
s
_
= arctan
_
:
_
A
2
1
2
__
. (4.27)
Finally the rotor rotating mmf wave is given by integration of the sinusoidal
current sheet, so it will lag the current sheet by 90

. This means that the
rotating rotor ‡ux-density wave, which is driven directly by the rotor mmf wave,
will lag the current sheet by 90

.
We conclude that
c
rs
= (90

+c) (4.28)
is the angle between the rotor rotating ‡ux-density (1…eld) wave and the
stator rotating ‡ux-density (1…eld) wave, upon which an incidental sign or
polarity change is overlaid. The angle c
rs
follows by accumulation of the phase
shifts which are incidental to the process described above. Hence when slip : is
small
c
rs
90

. (4.29)
Angle c
rs
becomes more negative than 90

if leakage reactance A
2
and/or
slip : increase. In cases where the resultant air-gap rotating ‡ux density wave
is approximately constant
c
r
c
rs
, (4.30)
where the torque angle c
r
is the angle between the rotor rotating ‡ux-density
wave and that resultant wave.
Refer to Figs. 6.5 and 6.6 which are pertinent to torque angle considerations
for wound rotors and squirrel-cage rotors, respectively.
4.8. TORQUE 99
4.8 Torque
Torque was derived from coenergy based on the magnetic …eld in the gap, in
Section 1.18.7 of previous notes entitled Rotating Machines. The derivation
was based on the assumption that the stator and rotor ‡ux-density waves ro-
tate in synchronism with each other, which is true for induction machines as
demonstrated above. This led to the expression for torque:
T =
_
¬
2
_
_
·
poles
2
_
2

sr
1
r
sinc
r
, (4.31)
which is equation 1.99 in the aforementioned notes (same as textbook equation
4.81). In steady-state operation the torque angle c
r
is constant so the torque
will be steady or constant in an induction machine.
Equation 4.31 is applied to an induction machine by assuming: the magni-
tude
sr
of the resultant air gap ‡ux
sr
is approximately constant for a given
applied stator voltage and frequency; and rotor mmf 1
r
is proportional to rotor
current 1
r
. Hence we can rewrite this equation as
T = 11
r
sinc
r
, (4.32)
where 1 is a constant, which gives
T = 11
r
(4.33)
by equation 4.29 if slip is small. E.g., under normal operating conditions slip
is 2% to 10% in a typical squirrel-cage motor, which means )
r
is in the range
of 1 Hz to 6 Hz for a 60 Hz motor. In such region 7
2
is predominately resistive
so 1
r
is increasing with slip because induced rotor voltage magnitude
¸
¸
¸
´
1
0
2
¸
¸
¸ is
increasing with slip. This produces a steep nearly linear rise starting at zero
slip in the torque vs. speed curve , as illustrated in Fig. 6.4.
Torque will reach a maximum (called breakdown torque) and then decrease
as slip increases further for two reasons, based on equation 4.32. The increase in
1
r
abates as 7
2
asymptotically approaches a constant reactance. Likewise sinc
r
decreases as the phase angle of 7
2
increases due to its reactive component, in
view of equations 4.28 and 4.30.
4.9 Analysis by Equivalent Circuit
4.9.1 Power and E¢ciency
Total (real) power transferred across the gap is
1
gap
= :
ph
1
2
2
_
1
2
:
_
(4.34)
100 CHAPTER 4. POLYPHASE INDUCTION MACHINES
with reference to the equivalent circuit in Fig. 6.9, because this is the power
transferred through the terminals a-b where the rotor equivalent circuit appears
with reference to the stator. The power dissipation in the rotor resistance is
1
rotor
= :
ph
1
2
2
1
2
(4.35)
which can be deduced either by referring 1
2
and 1
2
to the rotor side or by ob-
servation that equation 4.34 reduces to this expression is the rotor is stationary
(: = 1).
The mechanical power developed by the motor is
1
mech
= 1
gap
1
rotor
= :
ph
1
2
2
_
1
2
:
_
:
ph
1
2
2
1
2
(4.36)
= :
ph
1
2
2
_
1
2
:
1
2
_
= :
ph
1
2
2
1
2
_
1 :
:
_
.
This gives
1
mech
= (1 :) 1
gap
(4.37)
and
1
rotor
= :1
gap
(4.38)
by comparison with equations 4.34 and 4.35, respectively. Hence the fractional
power e¢ciency is
j
.
=
1
mech
1
gap
= 1 : (4.39)
so it is very ine¢cient to operate an induction motor at large slip :. Noting that
1
2
:
1
2
= 1
2
_
1 :
:
_
(4.40)
or
1
2
:
= 1
2
+1
2
_
1 :
:
_
(4.41)
from equation 4.36, it is useful to express
R2
s
by equation 4.41 for power analysis
as shown in the alternative equivalent circuit of Fig. 6.10.
Refer to Example 6.1 and Practice Problem 6.1.
4.9.2 Torque
Mechanical power is
1
mech
= .
m
T
mech
(4.42)
where T
mech
is mechanical torque, so
T
mech
=
1
mech
.
m
=
1
mech
(1 :) .
s
(4.43)
4.10. APPLICATION OF THEVENIN EQUIVALENT CIRCUITS 101
where .
m
= (1 :) .
s
. This gives
T
mech
=
1
gap
.
s
=
:
ph
1
2
2
_
R2
s
_
.
s
(4.44)
by application of equation 4.37, then equation 4.34; recall that
.
s
=
4¬)
e
·
poles
=
_
2
·
poles
_
.
e
.
Actual power and torque available at the mechanical shaft must take friction,
windage, and other mechanical losses into account. I.e.,
1
shaft
= 1
mech
1
rot
,
where 1
rot
represents such mechanical power losses, and
T
shaft
=
1
shaft
.
m
= T
mech
T
rot
,
where T
rot
represents such mechanical torque losses.
4.9.3 Examples
Refer to Example 6.2 and Practice Problem 6.2
4.10 Application of Thevenin Equivalent Circuits
4.10.1 Reduction of Motor Equivalent Circuit
´
1
2
must be evaluated in the equivalent circuit of Fig. 6.9 or Fig 6.10 to determine
characteristics such as torque and power vs. parameters such as :, 1
2
, and A
2
.
This evaluation can be simpli…ed, with additional insight as well, by substitution
of a Thevenin equivalent for the network to the left of terminals a-b in Fig. 6.9.
Recall the Thevenin equivalent from circuit theory. It requires determination
(of "dead" network) seen looking back into the terminals a-b. By voltage divi-
sion, the open circuit voltage is
´
\
1;eq
=
´
\
1
_
,A
m
1
1
+, (A
1
+A
m
)
_
(4.45)
where in most induction motors negligible error will result by neglecting the
stator winding resistance 1
1
. The equivalent impedance is
´
7
1;eq
= (1
1
+,A
1
) k (1
c
k,A
m
) (4.46)
or
´
7
1;eq
=
,A
m
(1
1
+,A
1
)
1
1
+, (A
1
+A
m
)
(4.47)
102 CHAPTER 4. POLYPHASE INDUCTION MACHINES
where the input voltage source \
1
is replaced by a short circuit and 1
c
has been
neglected in the last equation.
Now
´
1
2
is evaluated by application of the Thevenin equivalent:
´
1
2
=
´
\
1;eq
´
7
1;eq
+,A
2
+
R2
s
. (4.48)
4.10.2 Torque
Torque is
T
mech
=
_
1
.
s
_
_
:
ph
\
2
1;eq
_
R2
s
_
_
1
1;eq
+
_
R2
s
__
2
+ (A
1;eq
+A
2
)
2
_
(4.49)
from equation 4.44 with substitution for
´
1
2
from equation 4.48. Refer to Fig.
6.14 which shows a general plot of this expression for constant input voltage
and frequency. Refer to Fig. 6.15 which shows torque, power, and current for
the motor in Examples 6.2 and 6.3.
Maximum torque (breakdown torque) follows from equation 4.44 by …nding
the value of the resistance
R2
s
for maximum power transfer from the remainder
of the network. It is well known in network theory that this maximum occurs if
and only if the resistance equals the magnitude of the impedance seen looking
into the remainder of the network. Hence we take
1
2
:
=
_
1
2
1;eq
+ (A
1;eq
+A
2
)
2
(4.50)
and solve for
:
max T
=
1
2
_
1
2
1;eq
+ (A
1;eq
+A
2
)
2
. (4.51)
Substitution of :
max T
into equation 4.49 yields the maximum torque:
T
mech
=
_
1
.
s
_
_
_
0.5:
ph
\
2
1;eq
1
1;eq
+
_
1
2
1;eq
+ (A
1;eq
+A
2
)
2
_
_
. (4.52)
4.10.3 Examples
Refer to Examples 6.3 and 6.4.
4.11 Parameter Determination
4.11.1 Introduction
Motor parameters can be determined by a combination of no-load, blocked rotor,
and DC resistance measurements. No-load 1
2
1 losses in motors are higher than
4.11. PARAMETER DETERMINATION 103
1
2
1 losses in transformers under open circuit test conditions because motors
require greater excitation current due to the reluctance of the air gap. Stator
resistance should be measured separately with a DC ohmmeter. A blocked rotor
test is analogous to a short-circuit test of a transformer; the equivalent circuit
is identical.
Operating at no load, the motor voltage \
1
(line-to-neutral), current 1
1
, and
input power 1
nl
are measured. The rotational losses due to friction and windage
are
1
rot
= 1
nl
:
ph
1
2
1
1
1
(4.53)
where rotor 1
2
1 losses can be neglected because the rotor current is very small
at no load. Alternatively, …nd 1
rot
from power-o¤ decay transient:
J
d.
m
dt
= T
rot
=
1
rot
.
m
(4.54)
or
1
0
rot
= .
m
J
d.
m
dt
(4.55)
from physical laws. Hence abruptly switch o¤ power and measure
d!m
dt
to obtain
1
0
rot
which now does not include core losses. This allows determination of core
losses:
1
core
= 1
nl
1
0
rot
:
ph
1
2
1
1
1
. (4.56)
Also observe that
A
nl
A
11
= A
1
+A
m
(4.57)
because impedance of paralleled branch containing A
2
+
R2
s
is very large for
small :.
4.11.3 Blocked Rotor Test
Operating with blocked rotor, the motor voltage \
1
(line-to-neutral), current 1
1
,
and input power 1
bl
are measured. Voltage and frequency can be adjusted to
simulate actual operating conditions in the rotor due to slip. E.g., to simulate
: = 0.25 the test frequency should be 0.25)
e
and result at )
e
can be found by
backward extrapolation.
First determine blocked-rotor reactive power:
Q
bl
=
_
o
2
bl
1
2
bl
(4.58)
where
o
bl
= :
ph
\
1
1
1
(4.59)
Then the blocked-rotor reactance corrected to rated frequency )
r
is
A
bl
=
_
)
r
)
bl
__
Q
bl
:
ph
1
2
1
_
. (4.60)
104 CHAPTER 4. POLYPHASE INDUCTION MACHINES
The blocked-rotor resistance is
1
bl
=
1
bl
:
ph
1
2
1
. (4.61)
The equivalent circuit parameters can be determined by matching the mea-
surment
7
bl
= 1
bl
+,A
bl
(4.62)
to the input impedance of the equivalent circuit in Fig. 6.11 with : = 1. The
input impedance is
7
bl
= 1
1
+,A
1
+ (1
2
+,A
2
) k (,A
m
) (4.63)
= 1
1
+1
2
_
A
2
m
1
2
2
+ (A
m
+A
2
)
2
_
+,
_
A
1
+
A
m
_
1
2
2
+A
m
(A
m
+A
2
)
_
1
2
2
+ (A
m
+A
2
)
2
_
(4.64)
from the equivalent circuit. This reduces to
7
bl
= 1
1
+1
2
_
A
m
A
m
+A
2
_
2
+,
_
A
1
+A
2
_
A
m
A
m
+A
2
__
(4.65)
by taking approximations such as 1
2
A
m
.
Matching resistive and reactive components from this last equation to 7
bl
gives:
1
bl
= 1
1
+1
2
_
A
m
A
m
+A
2
_
2
(4.66)
and
A
bl
=
_
A
1
+A
2
_
A
m
A
m
+A
2
__
. (4.67)
Now these equations can be solved for
A
2
= (A
bl
A
1
)
_
A
m
A
m
+A
1
A
bl
_
(4.68)
and
1
2
= (1
bl
1
1
)
_
A
m
+A
2
A
m
_
2
. (4.69)
where A
1
+A
m
is known from the no-load test.
4.11.4 Examples
Refer to Examples 6.5 and 6.6.
Chapter 5
Single- and Two-Phase
Induction Motors
5.1 Introduction
Analysis of single- and two-phase induction motors is closely related to the
analysis given in the previous notes entitled Polyphase Induction Machines.
Recall that stator winding currents are balanced (unbalanced) if the net stator
mmf wave has only a forward-rotating (both forward- and backward- rotating)
component(s). The analysis for polyphase (three-phase) induction machines was
based on the assumption that stator currents are balanced. On the other hand
stator currents are inherently unbalanced in a single-phase motor and may be
unbalanced or balanced in a two-phase motor.
Analysis under unbalanced conditions is achieved by superposition of the
separate e¤ects of forward- and backward- rotating mmf waves. I.e., the e¤ect
of each wave is determined by the same method used to …nd the e¤ect of the
forward-rotating wave in the analysis of the polyphase induction machine. With
respect to the backward-rotating wave, it is only necessary to apply the fact
that the rotor slip becomes 2 :, where : is the rotor slip with respect to the
forward-rotating wave.
5.2 Single-Phase Induction Motor
5.2.1 Decomposition of mmf Wave
The fundamental sinusoidal mmf for a single phase winding is
F
a
(0) = 1
max
[cos 0
ae
] [cos .
e
t] (5.1)
where
i
a
= 1
a
cos .
e
t, (5.2)
105
106 CHAPTER 5. SINGLE- AND TWO-PHASE INDUCTION MOTORS
0
ae
.
=
_
·
poles
2
_
0
a
, (5.3)
and
1
max
.
=
_
4
¬
__
/
w
·
phase
·
poles
_
1
a
. (5.4)
Note that here F
a
(0) remains …xed in space with the envelope cos 0
ae
while its
amplitude is pulsating with respect to time due to the factor cos .
e
t. Refer to
Fig. 4.28.
The decomposition of F
a
(0) into the sum of forward and backward travelling
waves is accomplished simply by applying the trig identity
cos ccos , =
_
1
2
_
[cos (c ,) + cos (c +,)] (5.5)
to equation 5.1 where we take
c = 0
ae
(5.6)
and
, = .
e
t. (5.7)
This yields
F
a
= F
+
a
+F

a
(5.8)
where the forward-travelling wave is
F
+
a
.
=
_
1
max
2
_
[cos (0
ae
.
e
t)] (5.9)
and backward-travelling wave is
F

a
.
=
_
1
max
2
_
[cos (0
ae
+.
e
t)] . (5.10)
Here the subscript a refers to phase a and the superscripts + and refer to
forward- and backward- travelling waves, respectively. This decomposition of
the mmf wave due to a single phase was given in section 1.17.2 of the previous
notes entitled Rotating Machines; it is illustrated in phasor form by Fig. 4.28c.
5.2.2 Equivalent Circuit
The stator equivalent circuit for steady-state operation with a forward mmf
wave was derived in Sections 1.5 and 1.6 of the previous notes entitled Polyphase
Induction Machines; it is shown in Fig. 6.9. This equivalent circuit gives the
constraint between the stator voltage
´
\ and the stator current
´
1 which generates
the forward wave, where
´
\ is the sum of voltage induced by the forward wave
and the voltage drop across the stator leakage impedance. This was based on the
assumption that no backward wave was present because of balanced operation
of the polyphase machine. However now the peak mmf
_
Fmax
2
_
in equation 5.9
5.2. SINGLE-PHASE INDUCTION MOTOR 107
is one-half of the value used in the previous derivation, so
´
\ must be scaled
by the factor
1
2
. This is accomplished simply by scaling the impedance of the
stator equivalent circuit by the factor
1
2
.
The stator equivalent circuit for steady-state operation with a backward mmf
wave is essentially the same as that for a forward wave, with the exception that
the rotor slip : is replaced by 2 :. This new value of slip follows because
.
m
= (1 :) .
s
(5.11)
and the backward wave is rotating at speed .
s
, so the relative rotational rate
between the rotor and the backward wave is
.
m
(.
s
) = (1 :) .
s
(.
s
) = (2 :)
. ¸¸ .
slip
.
s
. (5.12)
Likewise the impedance of the equivalent circuit for the backward wave must
be scaled by the factor
1
2
.
The net equivalent circuit, based on the decomposition in equation 5.8 and
shown in Fig. 9.11c), follows from the scaled individual equivalent circuits for
the forward and backward waves. I.e., the portions of these individual equiv-
alent circuit impedances which represent voltages induced in the stator (pri-
mary) winding by forward and backward ‡ux-density waves are placed in series
because these induced voltages are additive in the winding, whereas the single
stator leakage impedance is common to both waves. The equivalent circuit in
Fig. 9.11c) indicates that the magnitude of the forward ‡ux-density wave will
increase relative to that of the backward ‡ux-density wave as slip increases,
because this leads to the relation
´
1
main;f

´
1
main;b
between the induced volt-
ages due to these waves. This occurs because the rotor currents induced by the
forward and backward ‡ux-density waves have the e¤ect of increasing and de-
creasing the forward and backward mmf waves, respectively. Hence the torque-
speed characteristic of the single-phase machine is more favorable than would
be expected from superposition of characteristics corresponding to constant for-
ward and backward ‡ux waves. I.e., the latter behavior would be equivalent
to superimposing separate forward and reverse torque-speed characteristics of a
polyphase induction motor driven by a constant voltage in each case.
The resultant torque-speed characteristic, illustrated in Fig. 9.2b), accounts
for the magnitude variation of forward and backward ‡ux-density waves with
slip. The magnitudes of these forward and backward waves can be determined
from
´
1
main;f
and
´
1
main;b
in the equivalent circuit of Fig. 9.11c). The forward
and backward torques corresponding to these waves (hence corresponding to
these induced voltages
´
1
main;f
and
´
1
main;b
) follow from the previous analysis of
polyphase induction machines. The resultant torque is the di¤erence of forward
and backward torques.
There is no starting torque in an induction motor with only one (main) stator
winding, as seen from the zero value of the resultant torque-speed characteristic
at .
m
= 0 (: = 1). This is expected because the single pulsating stationary
108 CHAPTER 5. SINGLE- AND TWO-PHASE INDUCTION MOTORS
spatial mmf distribution was resolved only mathematically into forward and
backward mmf waves. However a net torque is developed once the rotor is
moving forward because the e¤ect of the forward ‡ux-density wave becomes
larger than that of the backward ‡ux-density wave, due to the di¤erent rotor
slips relative to the forward and backward waves. The introduction of a second
stator winding to obtain starting torque is discussed in Section 5.4.
Physical insight is gained by reviewing the interpretation of the (general-
ized) transformer equivalent circuit which underlies the operational equivalent
circuits in Fig. 9.11c) for the forward- and backward- ‡ux density waves. Con-
sider the input current
´
1
main
which enters either of these operational equivalent
circuits; the forward equivalent circuit su¢ces for this discussion. All of current
´
1
main
would enter the magnetizing reactance 0.5A
m;main
(primary magnetizing
inductance) and produce a large self-inductive voltage
´
1
main;f
if the rotor (sec-
ondary) were open circuited. However the rotor loop impedance 0.5
_
A
2
+
R2
s
_
,
which has been referred to the stator (primary) side, shunts the magnetizing in-
ductance and diverts a signi…cant fraction of
´
1
main
away from this inductance.
This shunt current represents the fraction of
´
1
main
for which the attendant mag-
netic …eld is cancelled by the reaction …eld due to the induced rotor current.
This is an equivalent behavior with respect to the ‡ux-density wave (magnetic
…eld) whereas physically all of
´
1
main
‡ows through the magnetizing inductance
(primary).
5.3 Analysis by Equivalent Circuit
5.3.1 Forward and Backward Impedance
The forward and backward impedances are
7
f
.
= 1
f
+,A
f
=
_
1
2
:
+,A
2
_
k (,A
m
) (5.13)
and
7
b
.
= 1
b
+,A
b
=
_
1
2
2 :
+,A
2
_
k (,A
m
) , (5.14)
respectively. These are evident from Fig. 9.11c).
5.3.2 Power
The powers delivered by the stator winding to the forward and backward …elds
are
1
gap;f
= 1
2
(0.51
f
) (5.15)
and
1
gap;b
= 1
2
(0.51
b
), (5.16)
respectively. This follows from examination of Fig. 9.11c). A closer examination
will show that 1
gap;b
accounts for one-half of the power dissipation in 1
2
with
5.3. ANALYSIS BY EQUIVALENT CIRCUIT 109
stationary rotor (: = 1) and diminishes as : decreases. This occurs because the
"backwards" power "opposes" forward power.
5.3.3 Torque
The electromagnetic torques of the forward and backward ‡ux-density waves
(…elds) are
T
main;f
=
1
gap;f
.
s
(5.17)
and
T
main;b
=
1
gap;b
.
s
, (5.18)
respectively. Derivation of these results is exactly analogous to the derivation
of T
mech
in Section 1.9.2 of the previous notes entitled Polyphase Induction
Machines. The net torque is
T
mech
= T
main;f
T
main;b
=
1
.
s
(1
gap;f
1
gap;b
) (5.19)
because the torque of the backward …eld opposes that of the forward …eld.
Note that the interaction of the forward and backward waves will also produce
an opposing torque of zero average value which ‡uctuates at double frequency.
This undesirable e¤ect is one reason that the performance of polyphase induction
motors is better than that of single-phase induction motors.
5.3.4 Losses
The 1
2
1 losses due to forward and backward waves follow from the correspond-
ing portions of the equivalent circuit in Fig. 9.11c):
1
rotor;f
= :1
gap;f
(5.20)
and
1
rotor;b
= (2 :) 1
gap;b
, (5.21)
respectively. Derivation of these results is exactly analogous to the derivation
of 1
rotor
in Section 1.9.1 of the previous notes entitled Polyphase Induction
Machines; slip (2 :) plays the role of slip : in derivation of 1
rotor;b
. Finally,
the total rotor resistive power losses are additive because the rotor currents due
to forward and backward …elds are at di¤erent frequencies (orthogonality):
1
rotor
= 1
rotor;f
+1
rotor;b
= :1
gap;f
+ (2 :) 1
gap;b
(5.22)
5.3.5 Output and E¢ciency
The output mechanical power is
1
mech
= .
m
T
mech
= (1 :) .
s
T
mech
= (1 :) (1
gap;f
1
gap;b
)
110 CHAPTER 5. SINGLE- AND TWO-PHASE INDUCTION MOTORS
where the last expression follows by substitution of equation 5.19. The e¢ciency
is
j
.
=
1
mech
1
gap
=
(1 :) (1
gap;f
1
gap;b
)
1
gap;f
+1
gap;b
,
so it is ine¢cient to operate the motor at large slip.
5.4 Split-Phase and Two-Phase Motors
A second stator winding will create starting torque if the combined e¤ect of both
stator windings is to produce a forward ‡ux-density wave with greater magni-
tude than that of the backward ‡ux-density wave. Normally the distribution
of the second winding is in spatial quadrature with that of the main winding,
but the two windings may have di¤erent individual distributions and di¤erent
electrical characteristics (e.g., di¤erent resistance). To achieve the desired ef-
fect, the current in the second winding must be phase-shifted with respect to
the current in the main winding. Ideally the AC currents in these windings are

phase shift) but a smaller phase shift can work too.
An induction motor with two stator windings (as described above) may be
designed for connection to a single-phase voltage supply or designed for connec-
tion to a two-phase voltage supply. In the former case it is called a split-phase
motor because the required phase shift between winding currents must be ob-
tained by a di¤erence in the phase angles of the complex impedances presented
by the respective winding circuits. In the latter case it is called a two-phase
motor (or symmetrical two-phase motor if two identical windings are in spatial
quadrature). Phase shift in a two-phase motor results by connection to either
unbalanced or balanced (equal, quadrature) two-phase supply voltages. The an-
alytic approach is the same for split-phase and two-phase motors: decomposition
into forward and backward stator mmf waves. Analytically a split-phase motor
is equivalent to a two-phase motor operating under very unbalanced conditions.
5.5 Split-Phase Con…gurations
5.5.1 General
In a split-phase motor the second winding is called the auxiliary (or "start")
winding. The simplest method to obtain phase shift relative to the main wind-
ing is to increase the resistance of the auxiliary winding. E.g., the auxiliary
winding may be wound with …ner wire which can tolerate the required operat-
ing current long enough to start the motor. The auxiliary winding is switched
out automatically (e.g., by a centrifugal switch) after startup. Refer to Fig. 9.3
and Section 9.2.1 of the textbook.
5.6. BALANCED TWO-PHASE SYSTEMS 111
5.5.2 Capacitor-Type Motors
A capacitor is inserted in series the auxiliary winding to optimize the phase
shift (typically to 90

) between the main and auxiliary winding currents. This
achieves better starting torque than the simple scheme above. Refer to Section
9.2.2 of the textbook.
Capacitor-Start Motor
The series combination of winding and capacitor is switched out after start.
Refer to Fig. 9.4, Fig. 9.5, and Example 9.1.
Permanent Split-Capacitor Motor
The series combination of winding and capacitor is not switched out after start.
Rather this combination is used to improve the operating characteristics of the
motor at normal speed. E.g., torque pulsations can be reduced Refer to Fig.
9.6.
Capacitor-Start Capacitor-Run Motor
This resembles the permanent split-capacitor motor, except that the series ca-
pacitor is switched between two di¤erent values. One value optimizes starting
torque and the other value is optimal for operation at normal speed. Refer to
Fig. 9.7.
5.5.3 Other
Other motor types are shaded-pole induction motors, self-starting synchronous
reluctance motors, and hysteresis motors. These are not split-phase motors in
the same sense as the above but share some similar principles. The synchronous
reluctance motor only starts as an induction motor while the hysteresis motor is
not an induction motor. Refer to Sections 9.2.3, 9.2.4, and 9.2.5 of the textbook,
5.5.4 Examples
Refer to Example 9.2 and Practice Problem 9.2
5.6 Balanced Two-Phase Systems
The sinusoidal two-phase currents i
a
and i
b
(two-phase voltages ·
a
and ·
b
) are
called balanced only if they are equal in magnitude and 90

out of phase. This
de…nition derives from the context of a system where the corresponding identical
a and / stator windings are in spatial quadrature- i.e., at 90

orientation to
each other. Assume a reference coordinate system in which a positive angle
corresponds to counterclockwise rotation and for de…niteness the / winding axis
112 CHAPTER 5. SINGLE- AND TWO-PHASE INDUCTION MOTORS
is oriented at +90

with respect to the a winding axis. There are two cases for
balanced currents (balanced voltages): i
a
b
by 90

or i
a
lags i
b
by 90

a
b
by 90

or ·
a
lags ·
b
by 90

). The phase in the former case (latter case)
is called a positive sequence (negative sequence) because the resultant current
sheet hence mmf wave will rotate in the forward (backward) direction.
It su¢ces to prove the above claim for a positive phase sequence since the
result for a negative phase sequence will then follow by simply reversing the
a and / labels. For a postive sequence the superposition of the forward mmf
waves reinforces to produce a large forward wave and the superposition of the
backward mmf waves cancels to produce zero backward wave. The proof here
is analogous to that for a balanced three-phase system in Section 1.17 of the
previous notes entitled Rotating Machines. The balanced (positive sequence)
two-phase currents (in phasor notation) are:
1
a
= 1
m
\0

=)F
a
= F
+
a
+F

a
, (5.23)
1
b
= 1
m
\ 90

=)F
b
= F
+
b
+F

b
, (5.24)
where F
a
and F
b
denote their corresponding mmf waves.
The net mmf is one wave
F = F
a
+F
b
(5.25)
which can be decomposed into forward- and backward- travelling waves:
F = F
+
+F

. (5.26)
Clearly
F
+
= F
+
a
+F
+
b
(5.27)
and
F

= F

a
+F

b
. (5.28)
Analogous to Section 5.2.1, the individual forward-wave components are
F
+
a
.
=
_
1
max
2
_
[cos (0
ae
.
e
t)] (5.29)
and
F
+
b
.
=
_
1
max
2
_
[cos (0
ae
90

[.
e
t 90

])] =
_
1
max
2
_
[cos (0
ae
.
e
t)] ,
(5.30)
where the cosine arguments 0
ae
and 0
ae
90

are due to the spatial o¤sets of
windings a and /, respectively, and the cosine arguments .
e
t and .
e
t 90

are
due to the time (phase) o¤sets of i
a
and i
b
respectively. Hence
F
+
.
= (1
max
) [cos (0
ae
.
e
t)] (5.31)
from equation 5.27.
5.7. SYMMETRICAL COMPONENTS 113
Also analogous to Section 5.2.1, the individual backward-wave components
are
F

a
.
=
_
1
max
2
_
[cos (0
ae
+.
e
t)] (5.32)
and
F

b
.
=
_
1
max
2
_
[cos (0
ae
90

+ [.
e
t 90

])] =
_
1
max
2
_
[cos (0
ae
+.
e
t 180

)] (5.33)
=
_
1
max
2
_
[cos (0
ae
+.
e
t)] (5.34)
Hence
F

= 0 (5.35)
by equation 5.28.
We conclude that F = F
+
+F

= F
+
consists of only a forward-travelling
spatial mmf wave whose amplitude is twice that of the forward-travelling wave
due to a single phase acting alone.
Theorem 1 The forward-travelling wave rotates at the synchronous angular
velocity
.
s
.
=
_
2
·
poles
_
.
e
. (5.36)
Proof. At a given point on the advancing wave
0
ae
.
e
t = constant
or
_
·
poles
2
_
0
a
.
e
t = constant.
For a time interval 4t this gives
_
·
poles
2
_
40
a
.
e
4t = 0
or
40
a
4t
=
.
e
_
N
poles
2
_.
This last expression is the actual rate of angular advance and agrees with .
s
.
5.7 Symmetrical Components
The symmetrical component method is a decomposition of a pair of arbitrary
two-phase currents i
a
and i
b
into the sum of two balanced current pairs- a pair
with positive phase sequence and a pair with negative phase sequence. These
two balanced pairs or "components" are symmetrical in the sense that their
114 CHAPTER 5. SINGLE- AND TWO-PHASE INDUCTION MOTORS
equivalent current sheets are revolving in opposite directions with respect to
the horizontal axis. (Note that a similar development is possible for three-phase
currents.)
In phasor form, the decomposition implies
´
1
a
=
´
1
F
+
´
1
B
(5.37)
and
´
1
b
= ,
´
1
F
+,
´
1
B
, (5.38)
where
´
1
f
, ,
´
1
f
represents the positive-sequence balanced currents and
´
1
b
, ,
´
1
b
represents the negative-sequence balanced currents. In matrix form these equa-
tions are:
_
´
1
a
´
1
b
_
=
_
1 1
, ,
_
_
´
1
F
´
1
B
_
(5.39)
where the solution is
_
´
1
F
´
1
B
_
=
_
1 1
, ,
_
1
_
´
1
a
´
1
b
_
. (5.40)
Finally
_
´
1
F
´
1
B
_
=
_
1
2
__
1 ,
1 ,
_
_
´
1
a
´
1
b
_
(5.41)
by evaluation of the matrix inverse. This method can be applied to voltages as
well as currents, in which case the results will be identical in form. I.e., simply
substitute \ for 1 in the above equations. Refer to Figs. 9.12, 9.13, and 9.14.
5.8 Application of Symmetrical Components
Symmetrical two-phase motors can be analyzed by symmetrical components, as
follows. Arbitrary applied voltages
´
\
a
and
´
\
b
are decomposed into symmetrical
components
´
\
F
and
´
\
B
. The motor is analyzed once with only the balanced
pair
´
\
F
, ,
´
\
F
applied; this produces a balanced pair of stator winding currents
and only a forward mmf wave. Then the motor is analyzed again with only the
balanced pair
´
\
B
, ,
´
\
B
applied; this produces another balanced pair of stator
winding currents and only a backward mmf wave. Finally, the response to the
original input voltages is determined by superposition of individual responses
due to the forward and backward waves. Refer to Example 9.3.
5.9 Unsymmetrical Two-Phase Induction Mo-
tor
This general case is analyzed by application of the machine inductance matrix;
refer to Section 9.4.2 of the textbook. Note that the symmetrical component
5.9. UNSYMMETRICAL TWO-PHASE INDUCTION MOTOR 115
method cannot be applied directly in the unsymmetrical case because balanced
rotor current pairs do not follow from decomposition of applied voltages into
balanced pairs. E.g., a split-phase induction motor is unsymmetrical.
Refer to Examples 9.4 and 9.5.
116
Chapter 6
DC Machines
6.1 Introduction
DC machines have some desirable features:
1. versatility;
2. economy;
3. relatively simple drive systems;
4. ease of control (precise output);
5. wide range of motor speeds;
6. wide variety of volt-ampere or torque-speed characteristics;
This wide variety of characteristics is possible under steady-state or dynamic
operating conditions; it stems in part from the various possible operating …eld
con…gurations. I.e., the …eld winding(s) of the machine can be shunt-, series-,
compound-, or separately excited.
DC motors are and will continue to be widely applied even though AC ma-
chines have made some inroads due to the development of solid-state drives.
DC motor applications include automobile starter motors, traction motors for
subways and locomotives, and variable-speed electric drill motors.
Previously you received a link to a website where Class Notes 6 are posted
for DC Machines by Prof. Kirtley (MIT). Here we will refer to a few illustrations
from those notes; e.g., Kirtley- Fig. 7, etc.
6.2 Machine Construction
Refer to Fig. 7.1 for a diagram of a DC machine. The stator is salient pole
with a concentrated (…eld) winding, while the rotor is non-salient (cylindrical
117
118 CHAPTER 6. DC MACHINES
or "round") with a uniformly distributed (armature) winding. DC current is
supplied to the …eld and armature windings, where the connection to the rotat-
ing armature winding occurs by means of the commutator located on the rotor
shaft. The commutator functions as a mechanical switch or "recti…er" necessary
to maintain a spatially …xed DC current distribution (sheet) in the rotor. Refer
to Fig. 4.17 for an illustration of a simple (two-segment) commutator in an
elementary DC machine. Refer to Fig. 4.16 for a cutaway view of a DC motor
with multi-segment commutator in its right-hand side.
6.3 Stator BField
The operating 1 or ‡ux-density …eld in the air gap due to the concentrated
stator (…eld) winding is by design an approximation to a stationary square wave
as a function of the angle around the gap periphery. This ‡ux-density wave
alternates between positive and negative (‡at-top and ‡at-bottom) extremes,
as shown in Fig. 4.18a and reproduced by the short-dash curve in Fig. 7.11.
The ‡at tops and bottoms (constant sections) of the wave correspond to the
faces of the salient poles, where the air gap is uniform and small. The sloped
transition regions of the wave correspond to the large air gaps which occur
outside the arcs of the pole faces and these transitions are by design gradual
relative to those of a true square wave. These gradual transition regions are
desirable to reduce spurious e¤ects such as eddy currents, as explained later.
Note that the speed voltage induced in any turn of the armature winding will
have the same waveshape as the ‡ux-density wave in view of the ‡ux-cutting
principle.
6.4 Rotor
6.4.1 Armature Winding
The armature winding on a cylindrical rotor is shown schematically in Fig. 7.7a
for the speci…c case of a two-pole machine with 12 sections placed in corre-
sponding rotor slots. For analytic simplicity each section can be viewed as a
single loop or turn which can be enlarged to any number of turns with no loss of
generality. Loop conductors, indicated by small circles near the perimeter of the
rotor, run parallel to the long axis of the cylinder where currents entering and
leaving the page are indicated by crosses and dots, respectively. E.g., the …rst
loop consists of the outgoing (inner) conductor in slot 1, the dotted back-end
connection from slot 1 to slot 7, and the return (outer) conductor in slot 7. In
general each of the 12 loops are o¤set in sequence by 30

on the perimeter of
the rotor but are otherwise identical- consisting of an outgoing conductor and
a return conductor spaced at 180

.
The armature sections or loops are connected to establish a pair of DC
current sheets on the perimeter of the rotor, as shown in Fig. 1. These currents
sheets have odd symmetry about the vertical armature axis, since the right-hand
6.4. ROTOR 119
and left-hand current sheets enter and leave the page, respectively. The overall
connection of loops is shown in the developed form of Kirtley- Fig. 7, which
is essentially a series connection where commutator segments are attached as
periodic "taps". E.g., commutator segments 1 and 2 are attached to the outgoing
and return conductors of the …rst loop, respectively. A network equivalent
circuit for the armature in terms of lumped inductances is shown in Fig. 2.
6.4.2 Commutator
The commutator is shown schematically in the center of the rotor in Fig. 7.7a,
where the black rectangles indicate commutator brush contacts for access from
the external armature terminals. The commutator is shown in developed form
in Kirtley- Fig. 7. The numerous separate copper segments of the commutator
are imbedded at uniform intervals in insulating material around the rotor shaft.
As the rotor rotates, the commutation process is essentially the transfer of
one armature loop at a time from one current sheet to the other current sheet.
Referring to Fig. 7.7 where the rotor rotates counterclockwise, loop 1 is trans-
ferred from the right-hand to the left-hand current sheet while coincidentally the
diametrically opposed loop 7 is transferred from the left-hand to the right-hand
current sheet.
Consider the transfer of loop 1 which is equivalent to the transfer of 1
(1)
in
Fig. 2. Clearly the current in 1
(1)
after transfer must be equal and opposite
to that before transfer, so a transient voltage must appear across 1
(1)
because
·
(1)
= 1
(1) di
dt
. This occurs while loop 1 is brie‡y short-circuited as the commu-
tator brush spans segments 1 and 2, during which period a normal exponential
current decay occurs. However subsequently current must build in the opposite
direction as well, so an additional transient voltage is required. Refer to Fig.
7.8 which illustrates "linear commutation".
6.4.3 Armature BField
The armature (1 or ‡ux-density) …eld in the air gap is due to the aforemen-
tioned pair of current sheets. The mmf corresponding to these current sheets is
a stationary sawtooth wave as a function of the angle around the gap periphery,
as shown in the developed view of Fig. 4.23 and reproduced by the dash curve
in the developed view of Fig. 7.9. This mmf wave can be derived by applica-
tion of Ampere’s law with integration of the current sheet (density) within the
appropriate magnetic paths through the stator and rotor cores. Note that the
axis of symmetry of the current sheets, which lies at 90

to the direct axis, is
referred to as the quadrature axis of the machine. Maximum armature mmf
occurs on this axis.
The armature 1 …eld follows from the armature mmf wave with due regard
for the variation in gap as a function of the angle around the gap periphery, as
shown in Fig. 7.9. I.e., the armature 1 …eld is directly proportional to the
mmf wave in the region corresponding to the faces of the salient poles, where
the air gap is uniform and small. On the other hand the armature 1 …eld is
120 CHAPTER 6. DC MACHINES
greatly reduced in the region corresponding to the large air gaps which occur
outside the arcs of the pole faces.
6.4.4 Interpoles
The commutation process is susceptible to sparking or arcing due to the tran-
sient voltage required to reverse current in a loop. In small machines adjust-
ment of the brush position and linear commutation conditions can be su¢cient
to eliminate this problem. However in large machines additional measures may
be necessary.
Interpole windings are an additional measure which e¤ectively reverses a
loop current by creating an additional 1…eld in the neutral zone where the
stator …eld is zero. This additional …eld is placed where a loop is brie‡y short-
circuited by the brush, so the resulting induced voltage can drive the current
reversal through the brush. This is key to avoiding large transient voltages
because it drives the reversal during the short-circuit condition rather than
during the subsequent connection to the opposing current sheet. See Kirtley-
Fig. 9 and Fig. 7.21 for illustration of interpoles.
6.5 Speed Voltage
The net speed voltage which appears between a pair of diametrically opposed
commutator segments is the sum of the individual loop speed voltages because
these loops are essentially in series. Based on the approximation that
d
orig-
inates from a spatial ‡ux-density wave which consists of only a fundamental
sinusoid, the average value of the commutated (recti…ed) net speed voltage is
1
a
=

·
poles
C
a
2¬:

d
.
m
(6.1)
or
1
a
= 1
a

d
.
m
(6.2)
where 1
a
.
=
N
poles
Ca
2m
. This follows from equation 4.53 in the textbook where
C
a
.
= total number of conductors in armature winding, :
.
= number of parallel
paths through armature winding, and
d
is the direct-axis air-gap ‡ux per
pole. Note that the average value 1
a
is the same whether the C
a
armature
conductors are concentrated or distributed, but the ripple is greatly reduced if
the conductors are uniformly distributed. This is illustrated in Fig. 7.2
In the case where the spatial ‡ux-density is a ‡at-top wave as in Fig. 4.18b,
note that the individual loop speed voltages are also progressively shifted in
phase due to the angular displacement of the individual loops around the ro-
tor. Due to the commutator action, the net speed voltage seen at the brushes
approximates a constant (DC) waveform where the ripple is greatly reduced
due to the addition of these phase-shifted speed voltages. E.g., the commu-
tated (recti…ed) speed voltage for a single loop is shown in Fig. 4.18b, where
6.6. MAGNETIC FIELD ANALYSIS 121
there is a substantial ripple due to the original transition regions in the stator
1…eld. However for multiple phase-shifted loops the commutated sum will
have a waveform akin to that in Fig. 7.2 where sine rather than ‡at-top waves
are used for illustration. Here reduction of ripple is equivalent to reduction of
harmonic content in the Fourier series.
6.6 Magnetic Field Analysis
6.6.1 Resultant Field
The resultant 1…eld in the air gap is the sum of the …elds due to the stator
(…eld) winding and the rotor (armature) winding. This is illustrated in the
developed diagram of Fig. 7.11. A direct sum is taken because both the stator
and rotor …elds are basically radially directed in the gap.
6.6.2 Eddy Currents
Eddy currents will be induced in the rotor if it "sees" a changing magnetic …eld.
I.e., a changing magnetic …eld 1 induces an electric …eld 1 according one of
Maxwell’s equations (Faraday’s law in di¤erential form):
O 1 =
01
0t
. (6.3)
The electric …eld then causes circulating or eddy currents to ‡ow due to the
conductivity of the rotor core. This leads to i
2
1 losses so the core may be
laminated to reduce the e¤ect.
In the DC machine the resultant 1…eld wave is stationary while the rotor
turns through the wave. Hence
O 1 =

01
00

00
0t

(6.4)
from the above equation with
@B
@t
=

@B
@

@
@t

by the chain rule for di¤erenti-
ation. In other words the magnitude of the induced electric …eld (hence eddy
currents) is proportional to

@B
@

for the stationary resultant 1…eld wave,
given a speci…c mechanical rotor angular velocity .
.
=
@
@t
. Hence the stationary
stator 1 …eld wave is designed with the aforementioned gradual transitions
between its extreme ‡at-top and ‡at-bottom regions. Coincidentally the rotor
1…eld has inherently gradual transitions because it is a sawtooth wave. Also
eddy currents may not be a serious problem if the rotor angular velocity . is
su¢ciently small.
6.6.3 Cross-Magnetizing Armature Reaction
This form of armature reaction results from the interaction of the armature
…eld and the magnetically nonlinear core material. First consider the resultant
122 CHAPTER 6. DC MACHINES
1…eld shown in Fig. 7.11 for the ideal case where there is no armature reac-
tion. Here the …eld due to the armature mmf sums directly with that produced
by the stator winding, to produce a uniform tilt in the ‡at top of the resul-
tant ‡ux-density wave. However the net ‡ux per pole does not change and the
speed voltage does not change. The latter statement follows because the arma-
ture portion of the …eld moves with the armature so produces no net e¤ect in
Faraday’s law concerning rate of ‡ux change in a conducting loop.
Cross-magnetizing armature reaction occurs because nonlinear saturation
of the core material produces a non-uniform response to the additional mmf
created by the armature. In this case the resultant ‡ux-density wave is not the
simple sum of the individual waves; in other words the tilt in the ‡at top of
the resultant becomes non-uniform. Hence the net ‡ux per pole, torque, and
speed voltage will change. E¤ectively the armature mmf has acted to inhibit
the ability of the core material to respond to the stator mmf- a demagnetizing
e¤ect.
The e¤ect of cross-magnetizing armature reaction can be presented in a set
of magnetization curves where armature current is a parameter. This e¤ect
can be viewed in terms of an equivalent reduction in the e¤ective mmf available
from the …eld winding(s) hence an approximate translation of the magnetization
curve for zero armature current. Refer to Fig. 7.14 for magnetization curves.
6.6.4 Compensation Windings
Compensation windings are wound in the …eld pole faces to cancel cross-magnetizing
armature-reaction ‡ux in large machines. These windings mirror the armature
windings and are connected in series with them so as to be "self adjusting" with
respect to armature current. The net e¤ect is that only the ‡ux produced by the
…eld winding remains. Refer to Kirtley- Fig. 10 and Fig. 7.22 for illustrations
of compensation windings.
6.6.5 Torque
Torque was derived from coenergy of the air-gap magnetic …eld in Sections 1.18.6
and 1.18.7 of previous notes entitled Rotating Machines. That derivation was
based on the assumption that both the stator and rotor spatial ‡ux-density
waves are sinusoids of fundamental frequency and the most convenient form of
result for present use is:
T =

¬
2

·
poles
2

2

sr
1
r
sinc
r
. (6.5)
That derivation can be generalized to arbitrary stator and rotor spatial ‡ux-
density waves by using Fourier series representations for both. Generalization
yields a result which consists of the sum of many terms like the right-hand side of
equation 6.5, where
sr
, 1
r
, and sinc
r
in each term correspond to a speci…c har-
monic in the Fourier series. (It turns out that there will be no "cross-product"
6.7. POWER 123
terms between
sr
, 1
r
, and sinc
r
for di¤erent harmonics, due to the orthogo-
nality property.) Hence we can use equation 6.5 as an approximation for torque
in the present case where the ‡ux-density waves are not pure fundamental sinu-
soids. This amounts to dropping the harmonic terms from the full generalized
expression. Note that all even harmonic terms are zero anyway for a square
wave or sawtooth wave.
In the present case equation 6.5 specializes to
T =

¬
2

·
poles
2

2

d
1
a1
(6.6)
where
sr
=
d
is the direct-axis air-gap ‡ux per pole and 1
r
= 1
a1
is the
fundamental component of the spatial armature mmf wave. Note that sinc
r
1
because the angle c
r
between the armature and resultant mmf waves is nearly
90

. By Fourier series
1
a1
=

8
¬
2

(1
ag
)
peak
(6.7)
where
(1
ag
)
peak
=

C
a
2 / :·
poles

i
a
(6.8)
with C
a
.
= total number of conductors in armature winding, :
.
= number of
parallel paths through armature winding, and i
a
= armature current. Hence
T =

·
poles
C
a
2¬:

d
i
a
(6.9)
by substitution of equation 6.8 into equation 6.7 followed by substitution into
equation 6.6. We rewrite the last equation as
T = 1
a

d
i
a
(6.10)
where 1
a
.
=
N
poles
Ca
2m
.
6.7 Power
The mechanical power
1
mech
= .
m
(T) (6.11)
can be shown equal to the electromagnetic (gap) power 1
a
i
a
, by substitution of
.
m
=
1
a
1
a

d
(6.12)
from equation 6.2 and equation 6.10 into equation 6.11. This gives
1
mech
=

1
a
1
a

d

[1
a

d
i
a
] = 1
a
i
a
, (6.13)
124 CHAPTER 6. DC MACHINES
where rotational losses must be subtracted to determine power at the machine
shaft. The total electrical power (excluding series and/or shunt …eld windings)
at the machine terminals is
1 = 1
a
i
a
+i
2
a
1
a
where 1
a
.
= armature resistance.
6.8 Examples
Refer to Examples 7.1, 7.2, 7.3, 7.4, 7.5, 7.6, and 7.7.
6.9 Field Winding Connections
The …eld winding can be separately excited, connected in series, or connected in
parallel with the armature. It may also consist of a parallel (shunt) and a series
winding, which is called a compound connection. These various connections, il-
lustrated in Figs. 7.4 and 7.13, will produce a variety of machine characteristics,
as illustrated by Fig. 7.5 and 7.6 for generators and motors, respectively.
Analysis of machine characteristics for the various connections and parame-
ter values begins by imposing the connection contraints on the machine vari-
ables. E.g., for a series connection 1
f
= 1
a
, etc. Refer to Section 7.6 for analysis
of some generator and motor connections.

2

CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION – e.g.: solenoids; relays; electromagnets Continuous Energy-Conversion Equipment – for energy transfer in large quantities (continuous) – e.g.: motors; generators

1.3

Purposes of Analysis
understand process develop design/optimization methods for devices with speci…c requirements develop models for performance evaluation of system components

1.4

Scope of Present Treatment
Magnetic …elds only are considered for conversion medium. Transducers and force-producing devices are considered …rst. Continuous energy-conversion devices are considered later. For electric-…eld systems analytical methods are similar.

1.5

Force in Electric and Magnetic Fields (Lorentz)

In simultaneous electric and magnetic …elds, Lorentz force on a charged particle is F = q(E + v B) (1.1) where F = force (newtons) q = charge (coulombs) E = electric …eld strength (volts per meter) B = magnetic …eld strength (teslas) v = velocity of particle relative to magnetic …eld (meters per second). Note that superposition of electric and magnetic forces occurs. In a pure electric …eld B = 0; so by equation 1.1 the force is F = qE (1.2)

which acts in the direction of the electric …eld irrespective of any particle motion.

1.6. FORCE BY CHARGE AND CURRENT DENSITIES In a pure magnetic …eld E= 0; so by equation 1.1 the force is F = q(v B)

3

(1.3)

which acts in the direction determined by the vector cross product v B. Recall that the cross product v B de…nes a vector whose direction is perpendicular to the plane of vectors v and B and whose magnitude is jv Bj = jvj jBj sin ; (1.4)

: where = angle between vectors v and B. The sense of direction for v B follows from a right-hand screw oriented along the perpendicular. Rotate v into B to turn the screw, whereby the vector v B points in the direction of advance of the screw. Alternatively use the equivalent right-hand rule. I.e., point the thumb and index …nger in the directions of v and B; respectively. Then v B emanates outward from the palm.

1.6

Force by Charge and Current Densities

The following results apply on a "microscopic scale", meaning a volume V is very small compared to everyday dimensions but large compared to atomic dimensions. So " V ! 0" in a limit expression must be interpreted within these constraints rather than in the true mathematical sense. Theorem 1 FV = (E + v where : Fv = force density = force per unit volume (newtons per cubic meter) : = charge density (newtons per cubic meter). Proof. Consider a small element of volume V = A j lj ; (1.6) B) (1.5)

of cross sectional area A and length j lj as shown in Fig. 1, which contains charge q= V: (1.7) Assume that this elemental volume is moving with velocity v; so F = q(E + v B) = V (E + v B) (1.8)

by substitution of equation 1.7 into equation 1.1. This gives : FV =
V !0

lim

F = (E + v V

B):

(1.9)

14.4 CHAPTER 1.13) (1.6. The net force on a body can be determined by integrating FV over its volume.15.17) (1. Proof.14) where the last expression follows by equation 1. Remark 3 Equation 1. ELECTROMECHANICAL ENERGY CONVERSION Theorem 2 FV = J is the magnetic force density where J = current density (amperes per square meter). t (1.12) if V moves as part of a current in a continuous charge distribution : (In this discussion both i and l are vectors normal to the cross section A:) This volume transports charge q through a boundary surface with area A in time t.3. and the velocity is v= (1.10) It remains only to show that J = v where : i J= : A Now for the volume element V the charge density is : q = V so q= V = A j lj l t (1. The results in Theorems 1 and 2 are useful for analysis in cases where a large number of charged particles are in motion. This corresponds to a current i with magnitude A j lj : q jij = = .10 for magnetic force density corresponds to equation 1.16) where the last step before the arrow follows from equation 1. .15) (1. t t where the last expression follows by equation 1.11) B (1. Specialize Theorem 1 for E = 0 : FV = (v B) = ( v) B: (1. so A j lj : jij = = jJj = A A t j lj = jvj =) J = v.

Coil sides are length l = 0:3 m. Find torque on coil when coil current I = 10 A. from rotor axis. located at radius R = 0:05 m. This gives F = (I B0 ) l because I = JA (where I inherits the vector property of J with respect to direction).e. The moment or torque around the rotor axis is = 1+ 2 where 1 (jIj jB0 j l) x b =R F1 = F2 = and 2 = ( R) [jRj (jIj jB0 j l) sin ] z b with z a unit vector perpendicular to and out of the paper. Hence the integral reduces to F = V (FV ) == V (J B0 ) = lA (J B0 ) = (JA B0 ) l where V = lA is the volume of the wire with A as its cross-sectional area.2 ) A single-turn coil (rectangular loop on non-magnetic rotor) resides in a uniform magnetic …eld B0 = 0:02 T..1. FV is constant over a side because J has constant magnitude and direction over the length of the wire on that side and B0 is uniform. 114) Fig. 2 (RE: Fitzgerald (p. the forces F1 b and F2 are a couple which produces the net torque = which evaluates to = [2 jRj (jIj jB0 j l) sin ] z b [jRj (jIj jB0 j l) sin ] z b Note that the translational forces on the ends which close the loop cancel and also produce no torque around the rotor axix because both are parallel to it.1. For one side the force F is the integral of the magnetic force density vector FV over the volume of the wire.6.Fitzgerald et al) 5 See Fig. Solution 5 Apply equation 1. b b . Coil plane is inclined at angle with respect to plane surface normal to …eld. because the current I directed b into the paper and the vertical …eld B0 are perpendicular.10 to determine the force on each coil side. I. FORCE BY CHARGE AND CURRENT DENSITIES Example 4 (Example 3. [2 (0:05) (10) (0:02) (0:3) sin ] z = [ 0:006 sin ] z N m . By evaluation of the above equation for wire 1 F1 = (jIj jB0 j l) x b where x is a unit vector coincident with the x axis. Likewise for wire 2 F2 = because the current I is directed out of the paper. 3.

Here integration can be interpreted as a sum of in…nitesimal constituent forces.7 Magnetic Force in Electromechanical Devices All forces due to magnetic …elds in electromechanical energy conversion devices derive from the Lorenz force given by equation 1. In such cases the forces do not act only on elements carrying free current. The energy method is an alternative approach which can be used to determine the net external e¤ect of the detailed internal Lorenz force distribution. The direct integration is a straightforward problem only if B and J are simple functions. as in Example 3.) Likewise computation of J (e. bR a dV 0 : (1. superposition applies. i. In general the functions for the …eld B and current density J in a device component are not simple. 3 where a lossless magnetic energy storage medium appears. This integration requires a detailed knowledge of the magnetic …eld B and the current density J as functions of position.e. due to the stator currents in a motor) invokes a form such as the Biot-Savart equation: B= 0 4 Z V J (x0 . Note that any electrical and . Direct computation of such forces by means of this equation requires the integration of the force density Fv over the volume of some component in a conversion device.6 CHAPTER 1. This can be generalized to include multiple electrical and mechanical terminal pairs... to which a pair of electrical terminals and a pair of mechanical terminals are attached.. the B …eld computation (e. Clearly this approach is feasible only by means of numerical algorithms.1 where both are uniform. ELECTROMECHANICAL ENERGY CONVERSION 1. s assuming the displacement current is negligible. I. Finally there is the required integration of the force density Fv over the component volume. y 0 .10. It applies under the reasonable assumption that device components are rigid (nondeformable) bodies and enables some simplifying assumptions used for magnetic circuits. for the rotor currents in a motor) becomes complex.8 Energy Method The energy method derives from conservation of energy in a system where some energy is stored in a magnetic …eld.e. z 0 ) R2 (Note that this equation follows from the di¤erential form of Ampere’ law.18) 1. In such general cases. This often occurs by design to meet performance objectives. the magnetic …eld is part of an energy conversion device and couples the electrical and mechanical portions of the system. This is shown conceptually in Fig.. The energy method is presented in the next sectiion. In many situations results are su¢ ciently accurate and application is far simpler than the direct method. This situation may be due in part to the presence of necessary magnetic material which can be analyzed in terms of equivalent surface magnetization curents due to the net e¤ect of magnetic dipole alignment.g.g.

9 Magnetic Energy by State Variables The magnetic stored energy in the lossless medium of Fig.19. includes those which were necessarily extracted (mathematically) to yield the (ideal) lossless magnetic energy storage medium. (1.20) which is a simple algebraic rearrangement of equation 1. 3. Items 1 and 2 are represented by resistances connected to the electrical terminals. Of course loss mechanisms can simply be ignored if such does not introduce signi…cant errors.The term dWloss . 1. mechanical losses due to friction and windage. Hence it includes: 1. These losses (hence the term dWloss ) must be represented by electrical and/or mechanical elements external to the medium. Item 3 is represented by dampers connected to the mechanical terminals. Here we consider energy-conversion devices which consist of stationary and moving components so the magnetic circuit includes moving or varying air gaps. 3 follows by application of dWf ld = dWelec dWmech = id (ff ld ) dx.19 if all system energy storage is in magnetic …elds.19) where the following represent di¤ erential changes: : dWelec = electrical energy input = id = (ei) dt : dWmech = mechanical energy output = (ff ld ) dx : dWf ld = magnetic stored energy. In this case it is necessary only to add the term : dWloss = loss energy (converted to heat) to the right-hand side of equation 1. The conservation principle also applies to the entire system.1. In fact even mass (which is lossless) associated with a moving section of a magnetic core must be represented by an external mass element connected to the mechanical terminals. ohmic electrical losses due to winding currents and magnetic-core eddy currents. The energy conservation principle for the lossless magnetic storage medium is dWelec = dWmech + dWf ld (1.9. MAGNETIC ENERGY BY STATE VARIABLES 7 mechanical loss mechanisms must be extracted analytically from a real magnetic device to yield the lossless magnetic energy storage model. core magnetization losses due to hysteresis. 2. Here signi…cant energy is stored in the air-gap magnetic …eld which serves a "reservoir" between the . which accounts for all system loss mechanisms.

hence maintains = 0. x0 ) retains dependency on x0 because the current i( . Z Z ) Wf ld ( 0 . 4. because there is no magnetic force without a magnetic …eld. x0 ) is dependent on the …xed x0 as well as the varying during integration over the vertical path P2b . meaning that no knowledge of system history is necessary to determine the stored energy Wf ld ( 0 . x0 ) = dWf ld .) Equation 1. as shown in Fig.22) = dWf ld + dWf ld (1. 0 . x0 ) will be the same irrespective of the chosen path P: In other words increments of electrical and mechanical energy have the same e¤ect on the balance Wf ld ( . x): (1. x0 ) (1. (1. Magnetic energy storage is conservative (lossless) here so Wf ld ( 0 . where energy stored in gap.21 is greatly simpli…ed by choosing P as path 2 in Fig. So Z Z dWf ld = [id (ff ld ) dx] = 0. x0 ) = Z dWf ld = P2b Z 0 i( . and core …elds does not enter directly into the transformation process.21) P where P represents a path of integration from the initial point x = 0.. which reduces equation 1.20 is applied over the path P noting that i and ff ld are each functions of and x : i = i( . x) regardless of the order in which they are added or subtracted by varying and x: Hence and x are state variables. inductors and transformers). The evaluation of equation 1. x0 ) under the present conditions. 4. ELECTROMECHANICAL ENERGY CONVERSION electrical and mechanical portions of the system. Consider path P2a which starts with = 0.8 CHAPTER 1.24) P2a P2b Here the variations in and x are essentially independent over P2a and P2b . x0 )d : (1. this implies ff ld 0 also.25) P2a P2a by application of equation 1.24 to Wf ld ( 0 . . The magnetic stored energy is Z Wf ld ( 0 .g. = 0 to the …nal point x0 .23) The dependency on x enters due to moving device component(s). (1. (A path P can be de…ned by a function relation between and x.26) 0 Note that Wf ld ( 0 . leakage. This is in contrast to systems with …xed geometry (e.20. x) and ff ld = ff ld ( . maintains d = 0.

1.28) These results follow from …eld theory.e. MAGNETIC ENERGY BY STATE VARIABLES 9 Magnetic stored energy is also obtained by integration of the …eld energy density over the volume V of the …eld: Wf ld = Z V Z B H dB 0 0 ! dV: (1. .. B = H). this Wf ld = Z V B2 2 dV: (1.27) For soft magnetic material with constant permeability gives (i.9.

119. ELECTROMECHANICAL ENERGY CONVERSION Example 6 Electromagnetic Relay See Fig.4 ) The magnetic core B H characteristic is assumed linear and lossless for simplicity. Fig. magnetic core. x0 ). 5 (RE: Fitzgerald. we obtain the linear relation: = L (x) i: I. where x plays the role of gap length. x0 ) = . x0 ) and = Z 0 i( . e.p. 0 . and i: Mechanical terminal variables are displacement x and force ff ld produced by the magnetic …eld. 3.26: Wf ld ( 0 . by applying equation 1. permeability of magnetic material. in conformance with the requirements of Fig. Any friction loss due to armature pivot is extracted to external damper element connected to mechanical terminals. and armature comprise a lossless magnetic energy storage medium. 3. The resistance of the excitation coil is extracted to a lumped external resistance R: Mass of moving component of armature is extracted to external (lossless) mass element connected to mechanical terminals. x0 )d : 0 First solve the linear relation above for i : ) i( . By previous analysis of magnetic circuits.10 CHAPTER 1. Electrical terminal variables are . x0 ) = 0 1 d = L (x0 ) L (x0 ) Z 0 d = 0 2L (x0 ) : .e. L (x) is the inductance which depends only on geometry.. and armature position x: Solution 7 Now determine the stored magnetic energy as a function of x. )Excitation coil. L (x0 ) then substitute this into the equation above for Wf ld ( Z 0 which gives: 2 0 Wf ld ( 0 .

: : h = plunger height. MAGNETIC ENERGY BY STATE VARIABLES Example 8 (Example 3. 3. Substitution of this last equation into the previous equation gives N 2 ld 1 x d L (x) = 0 2g and by substitution of this equation for L (x) into the equation for Wf ld (i. x) = 2 2L (x) derived in previous example. 6 (RE: Fitzgerald.6 ) Electromagnetic Relay with movable plunger is shown. assume h g: Dimensions: d = 0:15 mm. l = 0:1 mm. x) : Wf ld (i.1. =) Calculate magnetic stored energy Wf ld as a function of plunger position (0 < x < d) : Solution 9 Use Wf ld ( . Core and plunger materials: permeability is assumed in…nite ( ! 1).9. N = 1000 turns. with the gap cross-sectional area Agap a function of x : x Agap = ld 1 d from geometry of structure.Fitzgerald et al) 11 See Fig. 122. but convert this to function of i (given here) by substitution of = (L (x)) i: This gives Wf ld (i. 2g 2 by previous analysis of magnetic circuits.. Fig. g = 2:0 mm. x) = where L (x) = 0N (L (x)) i2 2 Agap .p.2. x : d . MMF Data: i = 10 A. x) = " 0N 2 ld 1 2 (2g) x d # i2 : Finally evaluation for given parameter values yields " # 2 4 10 7 (1000) (0:1) (0:15) 2 Wf ld = (10) 1 2 (2 (0:02)) =) Wf ld = 236 1 x d J. g = air-gap length.

31 provides the means to deduce mechanical force ff ld from the energy function Wf ld ( .20 must yield the same dWf ld ( . Now applying equation 1. x) and can be converted to a function of i and x by insertion of the function relating to i: Evaluation of a partial derivative with respect to a variable requires that all remaining (in this case one) variables be held constant. Equation 1.30) i= @ x=constant and @Wf ld ( . x) (1. x).30 provides the means to deduce the current i from the energy function Wf ld ( . x) @x dx: =constant (1.31) ff ld = @x =constant Equation 1. x) = 2L (x) 2 by the results of Example 6. x) gives dWf ld ( . x) : (1. x) @ d + x=constant @Wf ld ( .31 to this gives: ff ld = and substitution of @ @x 2 2L (x) = =constant 2 [L (x)] 2 d [L (x)] .12 CHAPTER 1. Example 10 Any Linear Magnetic System By de…nition of a linear system = [L (x)] i and in this case the energy is 2 Wf ld ( .10 Magnetic Force from Energy Wf ld ( . This imposes the requirement that the coe¢ cients of d must be the same in both equations and likewise the coe¢ cients of dx must be the same in both equations. 2 dx which is a function of i. . Hence we conclude @Wf ld ( . x) = @Wf ld ( . However this is a purely mathematical requirement which imposes no requirement at all that such remaining variables be held constant during actual operation of a device. x) in terms of the di¤erentials d and dx: Note that this requirement holds for arbitrary values of d and dx because and x are independent variables. dx = [L (x)] i into this equation gives ff ld = i2 d [L (x)] .29) Both this equation and equation 1. ELECTROMECHANICAL ENERGY CONVERSION 1. Applying this concept to Wf ld ( . x) is already established as a function of the two (state) variables and x: The di¤erential of any function of two variables can be written in terms of its partial derivatives with respect to those variables.

e. 3. From this d [L (x)] = 4a1 x3 + 3a2 x2 + 2a3 x + a4 dx can be substituted into the …rst equation to obtain ff ld : See Fig. MAGNETIC FORCE FROM ENERGY Example 11 (Example 3.1. x Plot solenoid force at current 0:75 A. Hence it is necessary only to evaluate d[L(x)] from the given table by numerical approximation. 7 for plots: L (x).p. 126-27 for Matlab code. Fig. as function of position for range 0:2 1:8 cm. Solution 12 Apply i2 d [L (x)] . 124. 7 (RE: Fitzgerald. k = 1. we assume L (x) = a1 x4 + a2 x3 + a3 x2 + a4 x1 + a5 and poly…t generates the coe¢ cients ak . 5. :::. after the table data is entered. 2 dx ff ld = derived in Example 10.7 ) Solenoid Inductance=) Table given as a function of plunger position: x (cm) L (mH) 0 2:8 0:2 2:26 0:4 1:78 0:4 1:52 0:8 1:34 1:0 1:26 1:2 1:20 1:4 1:16 1:6 1:13 1:8 1:11 2:0 1:10 Note: x = 0 is position of full retraction. See textbook pp.3.10. 2. dx A fourth-order polynomial …t to the data for L (x) can be obtain by using the Matlab function poly…t: I.ff ld : .Fitzgerald et al) 13 See Fig.. where i = 0:75 A. is given here.

20.14 CHAPTER 1.33) 0 in place of equation 1. 3 become f ld and : We continue the analogy to Section 1.32) in place of equation 1.9 for systems with linear displacement. respectively. 0 )d : (1. ELECTROMECHANICAL ENERGY CONVERSION 1. Likewise i= and f ld @Wf ld ( . Now applying equation 1. Here torque f ld and angular displacement replace force ff ld and linear displacement x.30 and 1.35 to this gives: f ld = @ @ 2 2L ( ) = =constant d [L ( )] . ) @ (1.26.34) =constant = @Wf ld ( . Example 13 Any Linear Magnetic System with Rotational Motion (This example is analogous to Example 10) By de…nition of a linear system = [L ( )] i and in this case the energy is 2 Wf ld ( .31. The derivation of magnetic …eld energy in this rotational system is analogous to that in Section 1.11 Torque from Energy Consider a system with rotational rather than linear mechanical motion. ) = 2L ( ) analogous to the results of Example 6.9 to obtain Z 0 Wf ld ( 0 . . d 2 [L ( )] 2 2 and substitution of = [L ( )] i into this equation gives f ld = i2 d [L ( )] . 0 ) = i( . Likewise the mechanical terminal variables in Fig. ) @ (1. respectively. so the energy balance is dWf ld = dWelec dWmech = id ( f ld ) d (1.35) =constant replace equations 1. 2 d which is a function of i.

which is consistent with fact that L ( ) is the same for ! + 180 (rotor rotation through 180 ) Given parameters: L0 = 10:6 mH. 3. 15 where is measured between magnetic axis of stator coil and major axis of rotor. . Fig. Magnetic circuit consists of a single-coil stator and an oval rotor.4. Note second-harmonic variation of L ( ) . 8 (RE: Fitzgerald.p. Find torque as function of for current of 2 A: Solution 15 By the results of Example 13 = i2 d [L ( )] : 2 d f ld Here d [L ( )] d [L0 + L2 cos (2 )] = = d d so f ld 2L2 sin (2 ) = i2 ( 2L2 sin (2 )) : 2 Evaluation for given parameters: (2) 2 2 3 f ld ( )= 2 2:7 10 sin (2 ) =) f ld ( )= 1:08 10 3 sin (2 ) N m: Note that torque acts in direction to pull rotor axis into alignment with coil axis. This alignment maximizes coil inductance. Coil inductance L ( ) varies with rotor angular position : L ( ) = L0 + L2 cos (2 ) .Fitzgerald et al) See Fig.1. 128.9 ) Rotational system is shown.11. TORQUE FROM ENERGY Example 14 (Example 3. L2 = 2:7 mH.

ELECTROMECHANICAL ENERGY CONVERSION .16 CHAPTER 1.

1. TORQUE FROM ENERGY 17 .11.

18 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION .

11. TORQUE FROM ENERGY 19 .1.

The magnetic coenergy is : 0 Wf ld (i0 . I. (Henceforth subscripts of i and x will be omitted. where vs.38: 0 dWf ld (i. x) = i . Substitution of these last two expressions into the above equation gives 0 dWf ld (i. so 0 Wf ld (i.36) 0 which is the area below the curve in Fig. x) = di + (ff ld ) dx: is a (1.e. x) Note that d (i ) = id + di and dWf ld ( .e. as the meaning will remain clear aside from the evaluation of integrals. x0 ) = Z i0 (i. these two areas add to the area of the rectangle with upper corner at the point (i. x) will be independent of the path P 0 chosen in the i x plane.26 is the area above the same curve.37) i. x) = i Wf ld ( . 4 that led from equation 1. x) = id + di [id (ff ld ) dx] = di + (ff ld ) dx: In general 0 Wf ld (i.26. Next magnetic energy Wf ld is acquired by the …eld as the ‡ increases from 0 to 0 along path P2b while position is held ux constant at x0 : Fig. x) = id (ff ld ) dx (by equation 1.) Clearly 0 Wf ld (i. x0 ) di. Take the di¤erential of equation 1.38) The last equation can be reduced to a function of only i and x because function of i: Theorem 16 0 dWf ld (i.20 CHAPTER 1. i represents a general function or characteristic seen at the electrical terminalpair of the system. x) + Wf ld ( . Each characteristic vs. x) = 0 and it follows from equation 1.24 to 1. (1. ELECTROMECHANICAL ENERGY CONVERSION 1. x) = d (i ) dWf ld ( . 9. whereas the magnetic energy given by equation 1. (1..39 that the result Wf ld (i. we assume that the position x0 is attained by following path P2a before any magnetic energy is stored in the …eld.36 Z P0 0 dWf ld .12 Coenergy The magnetic coenergy (complement of the magnetic energy) is related to the magnetic energy on the path P2 in Fig. 9 corresponds to a …xed value of the position parameter x.39) Proof. 9 is a detailed illustration of the latter process. i shown in Fig. ) . x) : (1.20). Note that the result in equation 1.

1. for hard magnetic material (where H = Hc gives B = 0) as in permanent magnets.13 0 Magnetic Force from Coenergy Wf ld (i.44) @i x=constant and ff ld @Wf ld (i.41) Wf ld = 2 V Equation 1. The choice to use energy or coenergy is primarily based on mathematical convenience in the context of a given problem. MAGNETIC FORCE FROM COENERGY 21 was obtained by choice of one such path. x) = di + dx: @i @x x=constant i=constant (1.13. This follows because energy and coenergy are zero when B = 0 which is equivalent to H = Hc . Equation 1.27 remains unchanged.e. hence " # ! " # ! 0 0 @Wf ld (i. x) = (1.45) Equation 1. whereas and x were taken as state variables for the energy. we obtain " # 0 @Wf ld (i.45 provides the means to deduce mechanical force ..9. x) is already established as a function of the two (state) variables i and x. These results follow from …eld theory.42) whereas equation 1. x) 0 dWf ld (i. Hence i and x are state variables for the coenergy. this gives Z H2 0 dV: (1. (1.40 must be modi…ed to Z 0 Wf ld = V Z H B dH 0 Hc ! dV.40 as the special case (soft material) where Hc = 0: 1. x). The general reasoning process here is analogous to that in Section 1. Magnetic coenergy is also obtained by integration of the …eld energy density over the volume V of the …eld: ! Z Z H 0 0 (1.43) Matching coe¢ cients of di and dx in this equation with those in equation 1.44 provides the means to deduce the ‡ ux from the coenergy func0 tion Wf ld (i.40) Wf ld = B dH dV V 0 for soft magnetic material (where H = 0 gives B = 0). x) = @x " 0 # : i=constant (1. Equation 1.39.42 is general because it includes equation 1. With constant permeability (i. B = H). x) @Wf ld (i.

respectively. 0 ) = (i.22 CHAPTER 1. x) = id + di [id ( f ld ) d ] = di + ( f ld ) d : By analogy to Section 1. 0 ) di (1. Substitution of these last two expressions into the above equation gives 0 dWf ld (i. x) and can be converted to a function of and x by insertion of the function relating to i: The reasoning here is analogous to that in Section 1.47) replaces equation 1. ELECTROMECHANICAL ENERGY CONVERSION 0 ff ld from the coenergy function Wf ld (i. ) = di + ( f ld ) d : Proof.13 we obtain @Wf ld (i. ) (1. ) i= @i and f ld " 0 # (1. respectively. ) = id ( f ld ) d (by equation 1.36 and likewise Wf ld (i.44 and 1. Now torque f ld and angular displacement replace force ff ld and linear displacement x.48) =constant @Wf ld (i. ) = i 0 Wf ld ( . 3.12 for systems with linear displacement. Take the di¤erential of equation 1. ) = d (i ) dWf ld ( .46) 0 in place of equation 1.45. The derivation of magnetic …eld coenergy in this rotational system is analogous to that in Section 1.49) i=constant in place of equations 1. and likewise become the mechanical terminal variables in Fig.47: 0 dWf ld (i. ) Note that d (i ) = id + di and dWf ld ( .10 where more detail can be found.32). 1.14 Torque from Coenergy Consider a system with rotational rather than linear mechanical motion.38. So magnetic coenergy becomes Z i0 : 0 Wf ld (i0 . Then analogous to Theorem 16 we have Theorem 17 0 dWf ld (i. ) = @ " 0 # (1. .

2.Fitzgerald et al) For the relay of Example 3. x) = : 4g d .1. by application of equation 1.45: # " # " 0 1 @Wf ld (i.2 (which is Example 8 and Fig. …nd the force on the plunger if the coil is driven by a controller which produces a current i (x) = I0 x d A: Note that both the current i (x) and the force are functions of x: Based on the assumptions in Example 3. x) = i2 2 [L (x)] = i2 2 " 0N 2 ld 1 2g x d # : which can be written correctly as a function of x alone by substitution of i (x) : " # 2 I0 0 N 2 ld 1 x x 2 0 d Wf ld (i. 6 here) this is a magnetically linear system and the following result was obtained: L (x) = 0N 2 ld 1 2g x d : By application of the result from Example 18 ff ld = i2 2 dL (x) dx = i2 2 0N 2 l 2g : Force as a function of x is obtained by substitution of i (x) into the previous equation: 2 I0 0 N 2 l x 2 ) ff ld = : 4g d Also from Example 18 the coenergy here is 0 Wf ld (i.5. x) = [L (x)] i2 : (i.36: Z i0 Z i0 1 : 0 Wf ld (i0 .14. by equation 1. x) @ 2 [L (x)] i2 dL (x) i2 = = ff ld = @x @x 2 dx i=constant i=constant 23 : Example 19 (Example 3. TORQUE FROM COENERGY Example 18 Any Linear Magnetic System Here we repeat Example 10 by using coenergy instead of energy. x) di = [L (x)] idi = 0 2 0 0 The force (same result as obtained in Example 10) is. By de…nition of a linear system = [L (x)] i: In this case the coenergy is.

I.45 to this last expression would give an incorrect result ff ld (di¤erent from that found above) because partial differentiation with respect to x imposes the condition that the current i be held constant (at least mathematically).. this condition has been violated already by the substitution i (x) which renders i a function of x in the last expression. .24 CHAPTER 1.e. ELECTROMECHANICAL ENERGY CONVERSION However direct application of Equation 1.

10. x) @x " 0 : = =constant 4x !0 lim 4Wf ld ( . x) 4Wf ld ( . in view of the de…nitions of Wf ld and Wf ld : It is necessary only to show " # 0 4Wf ld (i. 10 by the two curves and the vertical line from point a to point c: These 0 two areas (hence j 4Wf ld ( 0 . it will be more convenient to use energy to determine force or torque if ‡ is given as a driving or regulated quantity. which moves the device operation onto the curve labelled x + 4x: The area enclosed between these two curves includes both j 4Wf ld ( 0 . x) 4x 0 (1. This equivalence of forces can be explained by consideration of the i characteristics of a typical device. x). .51 converge to the same quantity as 4x ! 0.50 and 1.51 the limit is taken while i = i0 is constant and is allowed to vary from 0 to 0 0 4 .50 the limit is taken while = 0 is constant and i is allowed to vary from i0 to i0 4i. as shown in Fig.1 Energy and Coenergy Considerations Equivalence of Forces Equations 1. which corresponds to bounding an area (hence bounding j 4Wf ld ( 0 . x) lim = lim . the right-hand expressions in equations 1.15. and curve segment bc: The ratio of this di¤erence amount to the remaining large bounded area goes to zero as 4x ! 0.e.50) =constant @Wf ld (i. x)j and 0 0 4Wf ld (i0 .51) respectively. x)j and 4Wf ld (i0 . As a general rule. line ac. x)j) in Fig. ENERGY AND COENERGY CONSIDERATIONS 25 1. x) = @x # : = i=constant " 4Wf ld (i. i=constant (1.15. as shown by subsequent examples. The establishes the claimed equality.1. The latter case is quite common. where the right-hand expressions have been added for utility.45 must yield the same result for the force ff ld when applied to the same device.. which corresponds to bounding an area (hence bounding 4Wf ld (i0 . x)) in Fig.52 follows by consideration of the conditions under which the partial derivatives are de…ned.15 1. These equations are repeated here for reference as ff ld = and ff ld @Wf ld ( . The equality in equation 1. 10 by the two curves and the horizontal line from point a to point b: For equation 1. For equation 1. (1. Likewise it will be ux more convenient to use coenergy to determine force or torque if current is given as a driving quantity. x)) di¤er by only the amount in the (roughly) triangular region bounded by the line ab. x) lim 4x !0 4x # .31 and 1. Note that a family of such characteristics can be generated by variation of the position parameter x: The device is assumed to be operating initially along the i curve shown and labelled for a particular position x: Then the position is changed from x to x + 4x.52) 4x !0 4x !0 4x 4x =constant i=constant i.

g.42.27 and equation 1.2 Direction of Forces The …eld force ff ld acts to decrease the magnetic energy at constant ‡ or ux increase the magnetic coenergy at constant current. from which force can be determined as described above. In the case of a single electrical terminal-pair. increase of coenergy at constant current implies that the force acts to increase the inductance of a device. respectively.. 10 shows that in general the energy and coenergy in a nonlinear system are not equal. . such systems may include complex geometry and/or very nonlinear magnetic materials. I. 1. where such derivatives must be determined by numerical approximation.15. The energy or coenergy method is still usable even if a general …eld solution is required.3 Linear Systems In a linear system = [L (x)] i. Thereupon forces are calculated by taking partial derivatives (as before). respectively. E. in this case the energy and coenergy can be computed by application of equation 1. examination of Fig.e.. E.53) Likewise it can be shown from …eld theory that energy and coenergy densities are equal: B2 = 2 H2 : 2 (1.40 or 1. commercial software is available to determine magnetic coenergy for a linear-displacement actuator as a function of displacement x. General solutions are obtained by numerical algorithms implemented in commercial software ("…eld solvers")..16 General Solutions A general solution for the magnetic …eld is necessary in systems where an equivalent magnetic circuit is inadequate or innaccurate.54) In a nonlinear system is not linearly proportional to i and equivalently B is not linearly proportional to H: Hence in that case the above relations do not hold.. ELECTROMECHANICAL ENERGY CONVERSION 1. 1. This follows directly by interpretation of equations 1. Also.g. I.26 CHAPTER 1. the force pulls device components in a way which would reduce the reluctance in the corresponding magnetic-circuit model. The …nite-element method is a common example of such algorithms.51.e.15.50 and 1. from which it follows by previous results that magnetic energy and coenergy are equal: 2 2L = Li2 : 2 (1.

. and Hsteel = Bsteel = 0 due to ! 1: Coenergy: h 0 i h 0 i 0 Wf ld = Wf ld + Wf ld ag steel where Wf ld (equation 1.e. This leaves 0 Wf ld = Wf ld h i ag = Z 2 0 Hag V 2 ! dV = 2 0 Hag 2 ! V where the integral is the application of equation 1. r1 = 2:5 cm: MMF (N i) Data: To be determined in solution (i.1. respectively. N and i are not speci…ed). By simple geometry V = 2gh (r1 + 0:5g) H2 . g = 3:0 mm.41 to the (overlapping) air 0 ag gap. g = air-gap length. in terms of dimensions and magnetic …eld in the two air gaps.Fitzgerald et al) 27 See Fig. r1 =radius of rotor: Dimensions: h = 1:8 mm. Bsteel = Bag is …nite.16. The integral reduces to the simple product of coenergy density 2 and volume V because Hag is taken as constant in the (overlapping) air gap. 11 (RE: Fitzgerald. GENERAL SOLUTIONS Example 20 (Example 3. Magnetic materials (stator and rotor): permeability is assumed in…nite ( ! 1). 135.p. : : : h = axial length (height perpendicular to page). h because 2 Hsteel 2 h 0 i ag and Wf ld h 0 i steel are found by integration of coenergy density i Wf ld 0 steel =0 = 0 2 Bsteel 2 = 0 in view of Hsteel = 0 (above).12 ) Magnetic circuit with stator and rotor free to turn about axis is shown.41) over the volumes of the air gap and the steel. a) In the (overlapping) air gap: Hag = Ni 2g because the net air gap length is 2g (due to two gaps in series). Fig. 3. Compute maximum torque. =) a) Derive an expression for torque acting on rotor.6. Neglect e¤ects of fringe …elds. =) b) Assume maximum ‡ density B = 1:65 T (to avoid excessive satuux ration of steel) in portion of air gap where rotor and stator overlap.

when 6= 0).e. which tends to align the rotor and stator pole faces. Its sign is positive so that it acts in a direction to increase the overlap angle . b) It is necessary to determine mmf (N i) to apply the above equation for torque.28 CHAPTER 1.49: " # 2 h @ =4 i3 5 i=constant f ld @Wf ld (i. Now evaluate above equation for torque: f ld = 4 10 7 (7860) 2 1:8 10 2 2:5 10 4 (3 [10 3 ]) 2 + 0:5 3 10 3 = 3:09 N m: N. ! ! ) Wf ld = 0 2 0 Hag 2 V = 2 0 Hag 2 (2gh (r1 + 0:5g) ) = 0 (N i) h (r1 + 0:5g) 4g 2 Torque follows by application of equation 1. First …nd Hag : Bag = 0 Hag =) Hag = Bag 0 = 1:65 = 1:31 106 A= m: 4 [10 7 ] Then from …rst equation in part a): N i = 2gHag = 2 3 10 3 1:31 106 = 7860 A turns. ELECTROMECHANICAL ENERGY CONVERSION closely approximates the volume of a cylindrical slice where radius r1 + 0:5g is taken midway in the gap and is given in radians.B.: Refer to Practice Problem 3. ) = @ 0 0 (N i) 2 h(r1 +0:5g) 4g i=constant @ = 0 (N i) h (r1 + 0:5g) : 4g 2 This is a constant torque which is present only when there is some overlap (i.6 (Fitzgerald et al) to determine torque h f ld h …rst deriving inductance L: Then f ld follows by application of f ld = i by i i2 2 dL( ) d : ..

) = i1 d 1 + i2 d 2 ( f ld ) d . analogous to equation 1.55.. 2 . ) by integration of equation 1.2 Methods of Analysis The energy and coenergy methods will be extended and applied here to systems with two electrical terminal-pairs.1.17. 12 undergoes rotary motion. The system state is determined by three independent variables because there are three terminal-pairs. E. A hybrid set is also possible.17.1 Multiply-Excited Magnetic Field Systems Background Electromechanical systems with multiple electrical terminals are quite common. =constant (1. A conceptual representation of a system with two electrical terminal-pairs and one mechanical terminal-pair is shown in Fig.17 1.17.17. and are the natural choice for state variables because they are the "running variables" to determine Wf ld ( 1 . The currents are given by i1 = and i2 = @Wf ld ( @ @Wf ld ( @ 1.57) analogous to equation 1. 2. @ 2.56) 1. This amounts to a generalization of the approach presented so far. Likewise most energy conversion devices require multiple electrical terminals to excite multiple windings which produce the magnetic …elds for intended operation. 1 2.55) for a system with two electrical terminal-pairs and one mechanical terminalpair. Angular displacement is usually chosen as the mechanical state variable. ) =constant 2 =constant (1. It is assumed that the mechanical portion of the system in Fig. 2 . 1. 12. ) 1 =constant 2 =constant (1.3 Analysis by Energy Method The di¤erential energy is dWf ld ( 1. 1. In this case 1 . (1.g. MULTIPLY-EXCITED MAGNETIC FIELD SYSTEMS 29 1. where analogous extension to more than two terminal pairs is also possible. Either ‡ linkages 1 and 2 or currents i1 and i2 may ux be chosen as electrical state variables.58) .32. The torque is given by f ld = @Wf ld ( 1 .30. for power (p = vi) measurement by a wattmeter it is necessary to obtain torque proportional to the product of two electrical variables (voltage v and current i). ) 1 =constant . 2 2.

) = i1 d 1 + i2 d 2 ( f ld ) d (by equation 1. The energy is Wf ld ( 10 . ) = 1 di1 + 2 di2 +( f ld ) d : (1. i2 . = 0) d 1 0 1. 13.61.38 for a system with one electrical terminal-pair.48 and 1. i2 . 20 .59) by integration of equation 1.59 because energy is a state function. 0 ) found by integration over any other path will be the same as that computed by equation 1.30 CHAPTER 1. and dWf ld ( 1 . ). 20 . by derivation analogous to that for equations 1.35. ) = d ( 1 di1 ) + d( 2 di2 ) dWf ld ( 1.60: 0 dWf ld (i1 . i2 . (1. 4 for the singly-excited case.49. = 0) d 0 Z + 2 10 i1 ( 1.62) 1 = @i1 =constant i2 =constant .4 Analysis by Coenergy Method 0 : Wf ld (i1 .61) Proof. Substitution of these last three expressions into the above equation gives 0 dWf ld (i1 . ) = The coenergy for a system with two electrical terminal-pairs is 1 i1 + 2 i2 Wf ld ( 1. d (i2 2 ) = i2 d 2 + 2 di2 . 2.17. this path begins by integrating over while 1 = 2 = 0 which contributes nothing because f ld = 0 when there is no magnetic …eld. 2. i2 . Z 20 i2 ( 1 = 0. Then this path completes by integrating over 2 while 1 = 0 and …nally integrating over 1 : The quantity Wf ld ( 10 . 2 = 20 . ) Note that d (i1 1 ) = i1 d 1 + 1 di1 .60) analogous to equation 1. i2 . 2 . 2.55 over the speci…c path shown in Fig. ELECTROMECHANICAL ENERGY CONVERSION analogous to equation 1. ) . that " # 0 @Wf ld (i1 . (1. It follows from equation 1.55). Take the di¤erential of equation 1. Theorem 21 0 dWf ld (i1 . 0) = (1. ) = i1 d 1 + = 1 di1 + 1 di1 + i2 d 2 + 2 di2 di2 + ( f ld ) d : 2 [i1 d 1 + i2 d 2 ( f ld ) d ] Remark 22 This theorem is analogous to Theorem 17. Analogous to the path in Fig.

63) and f ld = @Wf ld (i1 . i2 = 0. L12 . The coenergy is Wf ld (i10 .65) by integration of equation 1.64) Note that f ld is determined directly in terms of currents. i2 . L22 are self-inductances. = 0 ) di1 0 1. = 0 ) di2 + 0 Z i10 1 (i1 .5 Linear Systems 1 For a linear system = L11 i1 + L12 i2 = L21 i1 + L22 i2 . MULTIPLY-EXCITED MAGNETIC FIELD SYSTEMS @Wf ld (i1 . =constant i1 =constant (1.66) (1. ) @ # : i1 =constant i2 =constant (1. and all inductances are functions of angular position (in general). Further L21 = L12 due to the physical reciprocity of magnetic-…eld phenomena. L21 .68) 2 (1. L21 are mutual inductances.17. i2 . i2 . by application of equation 1.17. (1. 0 0) = (1. 20 .61. 1.6 Application of Energy Method to Linear Systems The immediate goal is to determine Wf ld ( 10 . The path of integration is analogous to that used to derive equation 1.17.1.70) L21 1 L22 1 L12 D + L11 D 2 (1. 0 ) for a linear system in terms of 10 . L12 . 20 .59.66 and 1. L22 . For this it is necessary to express i1 and i2 as functions of 1 and 2 which are the running variables of integration in equation 1. ) = @i2 " 0 31 2 " 0 # .59. 0 and the given inductance parameters L11 .69) .a signi…cant advantage. Z i20 2 (i1 = 0. Hence we …nd i1 and i2 by inversion of equations 1.59. i20 .67) and 2 where L11 .67: i1 = and i2 = where D = L11 L22 L12 L21 : (1.

i20 .64 to equation 1. = 0:3 cos : 3 . after direct substitution of equations 1. 15 (RE: Fitzgerald. It turns out that the coenergy method leads to much simpler expression of results. 0 and the given inductance parameters L11 . and 1.15.pp. after substitution of equations 1. Inductances are speci…ed directly: L11 L22 L12 = (3 + cos 2 ) 10 = 30 + 10 cos 2 .65. gives Z 10 Z 20 [L22 ( 0 )] 1 [L12 ( 0 )] 20 [L11 ( 0 )] 2 Wf ld ( 10 . 20 . 0 ) will generate a very complicated expression for the torque. 14.7 Application of Coenergy Method to Linear Systems 0 The immediate goal is to determine Wf ld (i10 . L21 . 0 ) for a linear system in terms of i10 .71) d 1 D ( 0) D ( 0) 0 0 = [L11 ( 0 )] 2D ( 0 ) 2 20 + [L22 ( 0 )] 2D ( 0 ) 2 10 [L12 ( 0 )] 10 D ( 0) 20 : The torque can be determined by application of equation 1.73) by direct application of equation 1. 0 ) = d 2+ (1.72.65. 20 . Example 23 (Example 3. by application of equation 1. 140-41. 3.59.69.Fitzgerald et al) See Figs.17.7. ) = 0 [L22 ( )] i2 [L11 ( )] i2 2 1 + + [L12 ( 0 )] i1 i2 : 2 2 (1.58. i2 . 3. 1. L22 .72) The torque is " # 0 @Wf ld (i1 .32 CHAPTER 1. which requires partial di¤erentiation of the above expression with respect to 0 . Unfortunately partial di¤erentiation of this somewhat complicated expression for Wf ld ( 10 . i20 . ) f ld = @ = i1 =constant i2 =constant i2 d [L22 ( )] i2 d [L11 ( )] d [L12 ( )] 1 + 2 +[i1 i2 ] 2 d 2 d d (1. 1. . i2 .68. L12 .67.66 and 1. System has two electrical terminal-pairs (rotor and stator windings) for excitation. Figs. in particular due to functions of 0 which are present in all numerators and denominators.70.16 ) Magnetic circuit with stator and rotor free to turn about axis is shown. Clearly the coenergy method is far more e¢ cient than the energy method in the case of linear systems and gives the torque directly in terms of currents. yields Wf ld (i1 . ELECTROMECHANICAL ENERGY CONVERSION Then evaluation of equation 1. Evaluation of equation 1.

.1 Systems with Permanent Magnets Background Here we consider an electromechanical system or device which includes a permanent magnet. The torque component [i1 i2 ] d [L12 ( )] = d [i1 i2 ] (0:3) sin is due to the mutual interaction between rotor and stator currents and is hence called the mutual interaction torque. hence align their magnetic …elds.. I. For simplicity we assume that the system as given has no electrical terminal-pairs but note that our method can be generalized to systems which include multiple electrical terminal-pairs. SYSTEMS WITH PERMANENT MAGNETS Excitation Currents are: i1 = 0:8 A. these torques tend to align the rotor axis with the stator axis.18. hence maximize self inductances.e.73. These torques occur because self inductances are a function of rotor position. i2 = 0:01 A: =) Find and plot torque f ld ( ) : Solution 24 Torque can be computed from equation 1. This is also dependent on the reluctance of the magnetic path and acts in a direction to maximize mutual inductance.: f ld 33 = = = i2 d [L11 ( )] i2 d [L22 ( )] d [L12 ( )] 1 + 2 + [i1 i2 ] 2 d 2 d d 2 2 i i1 2 10 3 sin 2 + 2 ( 20 sin 2 ) [i1 i2 ] (0:3) sin 2 2 1:64 10 3 sin 2 2:4 10 3 sin : Remark 25 The torque components i2 d [L11 ( )] i2 1 = 1 2 d 2 and 2 10 3 sin 2 i2 d [L22 ( )] i2 2 = 2 ( 20 sin 2 ) 2 d 2 are due to winding currents acting separately as in a singly excited system. They are called reluctance torques and act in a direction to maximize coenergy. So they are dependent on the reluctance of the magnetic path where both individual winding currents actually "see" the same magnetic path.18. 1.18 1.1. It acts in a direction to align the rotor and stator.

Hence Z Z 0 0 0 Wf ld (if = 0. 0 In the most di¢ cult cases a general …eld solution will yield Wf ld or Wf ld by application of equation 1. Then magnetic force ff ld can be determined by 0 application of equation 1.18. Note that in the absence of any electrical terminal pair 0 Wf ld and Wf ld will be functions of only displacement x and the system (including magnet) parameters. 17. but ff ld is actually the desired result.75) f (if . x = x0 ) = dWf ld = (1. as shown in Fig. At this point we procede to the integration of equation 1.. 1. where we consider the simple system in Fig.39 but so far there is no state variable i: Now there is only a state variable ff ld to be integrated with respect to x. It turns out that a "…ctitious" or auxiliary winding with an electrical terminal current expedites the analysis here. Either of these equations will integrate over volume the magnetic …eld created in the system by the permanent magnet.2 Coenergy Method with Magnetic Circuit We wish to retain the advantages that magnetic-circuit analysis provides in those systems which are su¢ ciently uniform or simple to permit its application without excessive error. x = x0 ) where if = 0 is equivalent to 0 removing the …ctitious winding. In addition we must account for the e¤ect of the permanent magnet in determination of the …eld by magnetic-circuit analysis.45 to Wf ld .34 CHAPTER 1.31 to Wf ld or equation 1. ELECTROMECHANICAL ENERGY CONVERSION We wish to apply the energy/coenergy method so need to determine same. 16 for de…niteness. This path goes from the point (if = If 0 . Hence we choose the coenergy method. 0 The …rst task is to determine Wf ld as a function of displacement x. I. x = 0) where the system …eld 0 (hence Wf ld ) is zero to the point (if = 0. in which case current is the natural choice for a state variable. x = x0 ) by moving on the path P1 between these two points through the (deliberately chosen) segments P1a and P1b : Z Z 0 0 0 ) Wf ld (if = 0.39 where the second state variable has become i = if . These issues are resolved by introduction of the …ctitious winding (a mathematical arti…ce) shown in Fig. with = f : The integration is performed over the path P1 in the if x plane. x = x0 ) = dWf ld + dWf ld : (1.e. for which current is the natural state variable and for which analysis of linear systems is most e¢ cient (as shown previously). Previously this was done by integration of equation 1. x = x0 ) dif P1b If 0 R . respectively. we seek Wf ld (if = 0.42. The absence of an electrical terminal pair actually simpli…es the partial derivatives in these last equations.27 or equation 1. Initially the winding current if is adjusted (analytically) to a value If 0 which cancels the magnetic …eld in the permanent magnet (hence reduces the …eld to zero in the entire system). 16.74) P1a P1b 0 Note that P1a dWf ld = 0 because f = 0 (as well as dif = 0) due to the null …eld condition.

jBr j = 0:94T.. neglect …eld fringing e¤ects : : x = plunger displacement (movable gap). W0 = 2:0 cm: 0 =) a) Derive coenergy Wf ld as a function of plunger position x: =) b) Derive force on plunger as a function of plunger position x: =) c) Evaluate force at x = 0 and x = 0:5 cm: Solution 27 a) Represent the DC magnetization curve (RE: Fitzgerald. 145.75 is completely general so it applies in cases where any magnetic material in the system (including the magnet itself) is nonlinear. g0 = 0:2 cm.8. but here the "o¤set" current If 0 was necessary to render the integral zero along the …rst path segment. 1.18. 18 (RE: Fitzgerald.p. g0 = …xed air-gap length. x) . Widths: Wm = 2:0 cm. MMF Data: Fictitious winding (Nf turns with current if ) is included for analytic purposes.p. SYSTEMS WITH PERMANENT MAGNETS 35 where the term ff ld dx does not appear here because dx 0 on path P1b : (Note that the development here was analogous to that which led from equation 1. D = depth of core and plunger/ Dimensions: d = 2:0 cm. Note R Hc with R = 1:05 0 . Flux continuity requires Bm Wm D = Bg Wg D = B0 W0 D 0 H0 0 Hg . Wg = 3:0 cm. where we note that neither Wf ld nor ff ld is zero.45 to Wf ld (if = 0. The magnetic force ff ld due to the permanent magnet alone follows by applica0 0 tion of equation 1. The other ‡ densities are ux Bg = in the movable gap and B0 = in the …xed gap. Hc = 0 that the apparent coercivity Hc adjusted for the slight downward bend in the B H curve is somewhat larger than the actual coercivity. Fig.19) of samarium-cobalt by the linear approximation Bm = 0 R Hm Hc = 0 0 R Hm + Br where Br = 712 kA/m.24 to equation 1. Core and plunger materials: permeability is assumed in…nite ( ! 1).26.) The result in equation 1.Fitzgerald et al) See Fig.19 ) Magnetic circuit including permanent magnet (samarium-cobalt) with movable plunger is shown. D = 3:0 cm. d = magnet length. Fig. Example 26 (Example 3.1. 3. 36.

x = x0 ) H d = c = Nf if 2 Br d : R Nf 3 = Z 0 0 Hc d Nf if = 4 2 [Nf Wm D] 4 h d + Wm R Nf if R 0 Hc d x Wg 0 d + Wm 2 + Wm D (Br d) R 0 2 R x Wg + g0 W0 i5 .. In a similar vein. 0 Hg W g = 0 H0 W 0 This is solved for the vector of unknowns Hm . Hg . H0 and six equations (where the ‡ continuity equation counts as two). Hg . 3 g0 W0 5 dif . 0 Hc W m = 0 Hg W g R Hm by substitution of the …rst and second equations into the …rst "half" of the fourth equation. B0 .e. In matrix terms the system is ux sparse so it is simplest to reduce to a three-by-three system by combining some equations. Bg . H0 by matrix inversion and Hm is substituted into the …rst equation to yield: Bm = The ‡ linkage is ux = Nf Wm DBm = [Nf Wm D] 4 0 by substitution of the second and third equations into the last "half" of the fourth equation.36 CHAPTER 1. The last three equations form the three-by-three system which can be written in matrix form: 2 32 3 2 3 0 0 Hm R Wm 0 Wg R Hc W m 4 5 4 Hg 5 = 4 5: 0 0 0 Wg 0 W0 d x g0 H0 Nf if R Nf if R 0 Hc d x Wg 0 d + Wm + g0 W0 : f 2 R Nf if R 0 Hc d x Wg 0 d + Wm + g0 W0 3 5 where f = 0 when if = If 0 Then the coenergy is 0 Wf ld (if = 0. I. ELECTROMECHANICAL ENERGY CONVERSION and by Ampere’ law (for mmf): s Nf if = Hm d + Hg x + H0 g0 : There are six unknowns Bm . Hm .

x = x0 ) .18. c) Evaluation of the above expression for ff ld by substitution of known parameters gives ff ld = 115 N at x = 0 cm and ff ld = 85:8 N at x = 0:5 cm: .45 to Wf ld (if = 0. SYSTEMS WITH PERMANENT MAGNETS by application of equation 1. 3 0 by application of equation 1. where the negative sign indicates that the force acts to decrease the movable gap by pulling the plunger inward. b) The force is 37 ff ld = 2 6 4 2 0 Wg h 2 2 Wm D (Br d) d + Wm R 0 x Wg + g0 W0 7 i2 5 .75.1.

hard magnetic material: Bm = R Hm Hc 0 (1. 2. 0 R (1. ELECTROMECHANICAL ENERGY CONVERSION 1.3 Equivalent of Magnet with Linear Material Claim 28 Consider a section of linear. a section of soft.38 CHAPTER 1. ux Proof.. this will result in same ‡ in the external magnetic circuit.76) with area A and length d: With respect to the external magnetic circuit which it faces. By Ampere’ law s Hm d + Fe = 0 (1. s .18.e. linear magnetic material with the same permeability (B = R H) and same geometry.77) I.78) where Fe represents the mmf developed at the "terminals" of the external magnetic circuit which the magnet faces. ) Hm = Fe : d (1. Remark 29 Note the analogy between the previous theorem for magnetic circuits and Thevenin’ theorem for electrical networks. This equa0 tion gives the same result as the previous equation because N i = Hc by d equation 1.79) The ‡ produced in the external magnetic circuit by the permanent magnet is ux = ABm = A h Hm Hc 0 R i Hm = Fe d = RA Hc 0 Fe d : Now consider the replacement: = RA Ni d Fe d by magnetic-circuit analysis with application of linear superposition.77 for the mmf of the equivalent winding. this section can be replaced by: 1. and an equivalent winding of Ni = Hc d ampere-turns.

Core and plunger materials: permeability is assumed in…nite ( ! 1).9. MMF Data: Excitation winding (N1 = 1500 turns with current i1 ) is included. Solution 31 a) Represent the DC magnetization curve (RE: Fitzgerald. where the two mmf sources are in series with the variable gap. and magnet reluctances x .1.p. 19b. 3.22 ) Magnetic circuit including permanent magnet (neodymium-iron-boron) with movable plunger is shown. neglect …eld fringing e¤ects : : x = plunger displacement (movable gap). 149. Apply where Br = R Hc with R = 1:05 0 . 36.19) of neodymium-iron-boron by the linear approximation B= 0 R H Hc = 0 0 RH + Br 940 kA/m. RW D respectively. The equivalent magnetic circuit after this replacement is shown in Fig. Fig. d = magnet length. SYSTEMS WITH PERMANENT MAGNETS Example 30 (Example 3. 0W D and d Rm = . g0 = 1 mm.Fitzgerald et al) 39 See Fig. D = depth of core and plunger/ Dimensions: d = 8 mm. g0 = …xed air-gap length. For i1 = 0 this is equivalent to a single-winding system driven by (N i)equiv where 2 (N i)equiv 0 Li2 Wf ld = = : 2 2 [Rx + R0 + Rm ] The force on the plunger is " # 0 2 (N i)equiv @Wf ld dRx ff ld = = 2 @x dx [Rx + R0 + Rm ] = iequiv =constan t 2 (N i)equiv ( 0 W1 D) [Rx + R0 + Rm ] 2 . Rx = 0 W1 D g0 R0 = .p. …xed gap. Fig. 19 (RE: Fitzgerald. 1. Hc = Claim 28 to replace magnet by section of linear material with permeability R and equivalent winding of Ni = Hc d = 0 9:4 105 8 10 3 = 7520 ampere-turns. W1 = 4:5 cm: =) a) Find x directed force on plunger when current i1 = 0: =) b) Find current i1 required to reduce plunger force to zero. D = 3:5 cm.jBr j = 1:25T.18. Widths: W = 4:0 cm.

b) This occurs by setting the net mmf to zero: (N i)equiv + N1 i1 = 0 which gives i1 = (N i)equiv N1 = 7520 = 5:01 A: 1500 The direction of the current cannot be determined because the direction of magnetization of the magnet is not given.40 CHAPTER 1. ELECTROMECHANICAL ENERGY CONVERSION which gives ff ld = 703 N upon substitution of known parameters. ux . This current must be applied in the direction which reduces net ‡ to zero.

SYSTEMS WITH PERMANENT MAGNETS 41 .18.1.

ELECTROMECHANICAL ENERGY CONVERSION .42 CHAPTER 1.

1.18. SYSTEMS WITH PERMANENT MAGNETS 43 .

ELECTROMECHANICAL ENERGY CONVERSION .44 CHAPTER 1.

44 .

2.2) (2.1) S Electromechanical energy conversion occurs when mechanical motion produces d d dt 6= 0. Rotation of a B …eld past windings. The key principle is electromagnetic induction (Faraday’ law): s e= where = and : = Z N B dS: (2. 2.2 General Rotating Machine The general machine (a generator or a motor) consists of a stator (stationary member) and a rotor (rotating member) where the stator partially or totally 45 . 3. ux where rotor has geometry designed for that purpose.3) d dt (2. Rotation of windings through a B …eld. varies periodically by any of these methods due to the cyclic nature of mechanical motion in a rotating machine. Variation of reluctance in the main magnetic ‡ path by rotor rotation. whereby a physical machine can be reduced to a simple mathematical model. In rotating machines there are three methods to obtain dt 6= 0 mechanically: 1.1 Introduction Certain basic principles are common to both AC and DC rotating machines.Chapter 2 Rotating Machines 2.

2 DC Machine The armature winding is on the rotor and the …eld winding is on the stator in most cases. The DC interface dt between excitation or output and the periodic ("AC") induced armature (speed) voltage is performed by a commutator on the rotor. Induction Machine The …eld winding is shorted and no external source is applied directly to it so no rotating electrical contacts are required. as shown in Fig.1 AC Machines The armature winding is on the stator and the …eld winding is on the rotor almost invariably. Variable Reluctance Machine There is no …eld winding. Synchronous Machine The …eld winding is excited by DC current and requires rotating electrical contacts (slip rings) to feed it. whether energy is electrical output (generator) or electrical input (motor). Stator and rotor cores usually consist of magnetic material to concentrate and shape the B …elds produced by their respective windings. The armature of this machine is . A varying B …eld is produced by a non-uniform air-gap reluctance with respect to rotor angle. 1. This corresponds to method 1 in Section 2.2. In a generator mechanical torque is applied and the aforementioned torque opposes rotation. The …eld winding produces the operating ‡ of the ux machine and usually carries DC current.2. 2. it is the underlying reason for placing the armature on the rotor. Time-varying armature (stator) currents are applied. as explained later. It may be replaced by permanent magnets in some machines. Rather AC induction excites it by generalized transformer action between the stator and rotor.1 to produce d 6= 0. which is a reversal of positions with respect to an AC machine. ROTATING MACHINES encloses the rotor.1 to produce d dt 6= 0. This corresponds to method 2 in Section 2.46 CHAPTER 2. An exception is the so-called brushless DC motor in which the rotor is a permanent magnet. The commutator switches current in synchronism with rotation. An armature winding is a set of interconnected coils through which the bulk of machine energy passes. Torque is produced by the interaction of stator and rotor magnetic …elds and acts to align them. In a motor electrical energy (armature current) is applied and the aforementioned torque aids rotation. 2.

Eddy currents may occur in both stator and rotor of variable reluctance machines. Refer to Figs.2. Brushless 2. Torque is produced by the tendency of these …elds to align.3 Stator and Rotor Cores Material for these cores is usually magnetically soft and highly permeable electrical steel.3 in textbook. hence at a …xed angular displacement with respect to each other. Permanent magnet (PM) 4. Synchronous 3.5 Similarity of Physical Principles The following machines are apparently quite di¤erent but share a common principle with regard to the interaction between stator and rotor …elds. Variable Reluctance 6. 2.5. 2.1 DC Machine Both the stator and rotor B …elds are …xed in space. Core losses due to hysteresis and eddy currents must be considered. Actually the stator …eld can be analyzed as a superposition of forward and backward rotating waves. 2.1.2 AC Induction Machine Both the stator and rotor B …elds rotate together in space and remain at a …xed angular displacement with respect to each other. 4. Hysteresis 7. so laminated construction is used to reduce eddy currents. The latter …eld is …xed due to commutator action. where . 2. with windings placed in core slots. DC 2. This core material increases coupling between windings and the B …eld strength (hence stored magnetic energy) in a machine. 4. It also allows the designer to shape and distribute the B …elds.4 Some Machine Type-Names 1. STATOR AND ROTOR CORES 47 on the stator and driven by electronic switching circuitry which emulates a commutator. Induction 5.5. 4.2.3.

and integrand with respect to changes in a parameter z: This is restated for evaluation at a speci…c value z = z0 to show that the derivative operator can be moved to precede the integral in the last term: dF (z) dz = z=z0 db (z) dF (z) = f (b (z) . In other words. z) and the limits a (z) and b (z) may be functions of a parameter z. in general the integrand f (x. 2. lower limit.7 Gap Field Distribution in Rotating Machines Clearly a gap between the rotor and the stator is necessary to permit rotor rotation. Torque is produced by the tendency of these …elds to align. is necessary to achieve the desired output voltage waveform in a generator or desired torque characteristic in a motor. the magnetization is spread to attain a more or less uniform degree of magnetic-dipole alignment throughout the core volume.48 CHAPTER 2. For a given energy in the gap volume. Analytically this magnetization can be represented by equivalent surface currents on the cores.5) which is essentially a superposition of e¤ects due to changes in the upper limit. z0 )+ z=z0 Z b(z0 ) @ f (x. Coincidentally a prescribed distribution of the B …eld in the gap. The B …eld produced by the …eld winding continues through the gap and into armature slots where it is accessible to the armature winding. just as in a DC motor. Given s ux F (z) = Z b(z) f (x. We have @f (x.4) a(z) where both the integrand f (x. z) dx (2. as previously described. with respect to the axis of the …eld winding.6) 2. z) + dz Z b(z) db (z) dz f (b (z0 ) .6 Di¤erentiation Rule for Integrals (Leibnitz) Leibnitz rule is stated here for reference because it will be used to apply Faraday’ law to ‡ integrals. z) dx: @z a(z0 ) (2. z) dx dz dz da (z) f (a (z) . Also. the distributed nature of the B …eld reduces the tendency toward magnetic saturation at points in the cores where B attains its extreme values. z) dx @z a(z) (2. It turns out that nearly all magnetic …eld energy is concentrated in the gap volume because the H …eld is very small in highly permeable cores. . z) may contain another integral. z0 ) dx z=z0 da (z) dz f (a (z0 ) . ROTATING MACHINES the backward wave is zero in a 3-phase machine but the same as the forward wave in a single-phase machine.

.4.g. A simple bar magnet has two poles hence represents a two-pole winding. Hence B is directed radially in the case of a uniform circular gap.2. by design the magnetic …elds of the stator and rotor will have the same number of poles to achieve optimal performance. can be taken as the aggregate of the outputs with respect to all the magnetic poles.. The number of poles (Npoles ) in B is by de…nition the number of maxima which occur in jBj during a complete (360 ) revolution on the circumference of the gap. In a machine with a large number of pole-pairs each pair "spans" a small arc and contributes a small or narrow segment of the B …eld in a prescribed manner. Note that around the circumference B undergoes more or less complete reversals in direction because a winding sets up circulating currents.8.. 4. Bar magnets can be curved into horseshoe magnets to form a four-pole or eight-pole machine as shown in Fig.e. radially directed outward or inward). and 4. Generally one-half subcycle of the B …eld occurs in each of these narrow segments. Further we assume that this B …eld may vary with angle (in cylindrical coordinates) around the circumference of the gap but does not vary with linear (axial) displacement along the rotor or stator. the …eld B is speci…ed by its magnitude and sense (i. An analogy can be drawn with bar magnets and horseshoe magnets. 2. by design of the winding distribution or pole face geometry. salient-pole rotor or stator with concentrated winding). Hence the machine output.7 for illustrations of two-pole and four-pole machines including spatial distributions of their B …elds. These principles apply to a B …eld irrespective of whether it is produced by current in a …eld winding or an armature winding. Npoles is determined only by the winding pattern if the gap is uniform (e. MULTIPOLE FIELD DISTRIBUTION 49 The B …eld is normal to the rotor and stator surfaces due to boundary conditions on the magnetic core materials. 4.. Npoles will be dependent primarily on the design of the pole faces if the gap is non-uniform (e. whether it be voltage in a generator or torque in a motor. Npoles is always an even number because no magnetic monopoles exist.g.5. Refer to Figs. 4. round rotor and round stator with distributed winding). 3.6. In most multipole machines. For analytic purposes we assume that the B …eld is essentially constant across the gap from rotor to stator if the gap is small relative to the radius of the rotor. At any position on the circumference speci…ed by angle .8 Multipole Field Distribution Consider the radially directed spatial B …eld around the circumference of the gap due to an instantaneous current in either a …eld winding or an armature winding.

2. Thus its upper and lower wires bound the cylindrical half-shell at angles and + .8) S Hence the induced voltage in path C is I e = E ds = C d dt and the ‡ linkage to an N turn winding on this path is ux : = so the induced voltage in the winding is e= N d d = : dt dt (2. (By conservation of ‡ ux.) The plane of the rectangular loop is oriented at an arbitrary angle with respect to the vertical as shown in Fig. (2. 2.10 to a rotor (armature) winding consisting of a single rectangular loop (N = 1) turning through a B …eld produced in the gap by the stator (…eld) winding. This choice of S expedites evaluation of the surface integral in equation 2.7) C S where S may be chosen as any surface for which the curve C is a boundary and the total ‡ is ux Z : = B da: (2.10) N (2. t ld 0 .9 Induced Voltage by Application of Faraday’ s Law I d dt Z d dt Recall Faraday’ law: s E ds = B da = (2. as shown in Fig.11) S . This gives : = Z + Z 0 B da = B . Here the "curve" C is taken as the rectangular dt loop. The curve C and this surface S are shown in Fig. respectively.50 CHAPTER 2. 2. The surface S is taken as the cylindrical half-shell whose base is the planar section enclosed by the rectangular loop and whose cyclindrically curved surface coincides with the circular path of the loop conductors around the gap. ROTATING MACHINES 2.8 because the gap B …eld is everywhere normal to this surface. the same result would be obtained for the surface integral by choosing S as the aforementioned planar section but its evaluation would be complex.1 to produce d 6= 0.9) We apply equation 2. This corresponds to method 1 in Section 2.

for a generator this can be taken as the output voltage: This coincides with the classic "cutting of ‡ ux" interpretation.15 applies directly.13) l [B ( (t0 ) + ) B ( (t0 ))] ! (t0 ) l v [B ( (t0 ) + ) B ( (t0 ))] {z } r | Lorentz force density Thus we conclude that the waveform espeed (t) is simply a scaled version of B ( ) where = !t. with relative to the rotor axis.12) C C C C: C t=t0 C }A 1 : where ! = d = v : Note that the Lorentz force density term appears in equation dt r 2. OPTIMAL GAP FIELD 51 where l is the length of the loop. see Example 4. Now the speed voltage espeed (t) induced dt in the stator can be determined simply by taking the rotating rotor as the spatial frame of reference.10 Optimal Gap Field Optimal operation of a rotating machine requires that the (Fourier) spectral content of the induced speed voltage waveform espeed (t) matches the spectral . yields 0 e (t0 ) = d dt = t=t0 The speed voltage can be expressed as espeed (t0 ) = = B B Bd lB B dt B @ | (tZ)+ 0 (t0 ) B {z 0 .12 because this produces the E …elds down the length of the wires. Application of Leibnitz rule from equation 2. 2.2.10 to a stator (armature) winding consisting of a single rectangular loop (N = 1) where a rotating B …eld is produced in the gap by the rotor (…eld) winding.4 in textbook. Analogously we can apply equation 2.15) (2. This corresponds to method 2 in Section 2. By design of machine windings the B …eld is normally anti-symmetric around the gap: B ( (t0 ) + ) = so espeed (t0 ) = l [2B ( (t0 ))] ! (t0 ) and …nally if ! is constant espeed (t) = !l [2B (!t)] : (2.1 to produce d 6= 0.14) B ( (t0 )) . where we now take z = t. for the integral over the surface of the cylindrical half-shell.t d 0 + [B ( (t0 ) + ) | transform er voltage term } sp eed voltage term d B ( (t0 ))] dt {z (2. (2.10.6. Now the stator winding appears to be rotating with respect to the rotor …eld and equation 2. Hence the …eld appears stationary in time and can be denoted by B ( ).

By equation 2. The last statement follows by application of the orthogonality principle to harmonics with respect to the fundamental component. ROTATING MACHINES content of the required armature (generator) output or applied armature (motor) input.11 2. the gap …eld waveform B ( ) has the same spectral content as that required in espeed (t)..11.1 AC Machines Electrical and Mechanical Frequency The spatial variation in the gap B …eld will be more rapid in machines with a greater number of poles: Npoles (2. By Kircho¤’ voltage s law (KVL).. In addition harmonics tend to introduce small periodic variations in ! m : 2. In an AC machine this specializes to the simple requirement that espeed (t) is a pure sinusoid.e. Assume that the machine is in normal steady-state operation.18) Npoles 2 !m .15. In a DC machine this must be interpreted by taking the intervening commutator into account. These current components are undesirable because they increase the RMS current value hence increase resistive (i2 R) losses and increase core eddy current losses (which are proportional to the square of frequency) but do not contribute to average electromechanical power transfer.e. ae ) and a (2. These harmonic components in the loop voltage will drive corresponding current components around the loop. any mismatch in spectral content between espeed (t) and the voltage source yields a voltage di¤erence which contains additional harmonic components with respect to ! m .19) 2 60 . So B ( ) should be a pure (fundamental) sinusoid in an AC machine or a square wave (i. its spectrum contains no harmonics.16 with respect to time. piecewise constant) in a DC machine. rotating at a constant angular velocity ! m : Consider the electrical circuit or loop formed by the armature winding and a voltage source of the required spectral content connected directly to the armature terminals. It follows that !e = hence fe = Npoles 2 fm .16) a ae = 2 where ae is the electrical angle or argument of the cyclic …eld B ( is the mechanical or spatial angle around the gap. so the armature itself actually sees or should provide a square wave. From this last equation it follows that Npoles RP M fe = Hz. (2. (2. i.17) by di¤erentiation of equation 2. The importance of the spectral content in espeed (t) hence B ( ) is explained as follows.52 CHAPTER 2.

2. AC MACHINES 53 where the term RP M converts mechanical revolutions per minute to fm which 60 is in revolutions per second. 2. the …eld (rotor) winding is short-circuited so that no external excitation need be applied to it. E. equals the electrical frequency of the voltage or current applied to its armature winding.11. AC rotor current follows by induction (generalized transformer action) from stator to rotor. the mechanical frequency of rotation. Solution 4 Npoles = 4.11. What RPM is required to obtain fe = 60 Hz ? Solution 2 From the above equation RP M = 2 2 (60fe ) = (60 60) = 3600: Npoles 2 Note that the spatial B …eld of this machine completes one full cycle per mechanical revolution. Express electrical frequency in terms of mechanical frequency. Example 3 Given a 4-pole single-phase machine. Example 1 Given a 2-pole single-phase machine. consider a two-pole machine (Npoles = 2) where ! m < ! e so that the rotor rotation continuously slips behind rotation of the stator B …eld.3 Induction Machines In induction machines the armature (stator) winding is the same as in synchronous machines.18 matches the frequency of the (possibly "in…nite") bus to which it is connected. Hence there will be a continuosly induced rotor current of frequency !s = !e !m (2.2 Synchronous Machines Synchronous machine operation is best explained by separate consideration of motors and generators.18 always occurs in multipole machines (both generators and motors). An AC motor is operating synchronously if fe . Note that fe is the electrical frequency of the speed voltage and fm is the mechanical frequency of rotation in a given machine. Note that the inherent "frequency conversion" given by equation 2. the electrical frequency of the speed voltage which it produces.20) .11.. Conversely an AC generator is operating synchronously if fm is such that fe given by equation 2. Then fm . is constrained or locked to the applied frequency by equation 2.18.g. On the other hand. so by equation fe = Npoles 2 fm = fe = 4 2 fm = 2fm 2.

. hence no induced rotor current. refer to Fig. These sets are displaced by 120 electrical degrees in the spatial pattern around the rotor or stator.) Salient-pole rotors are most suitable in this case from the standpoint of mechanical construction and economy.e. 2. (Electrical power systems operate at fe = 60 Hz or fe = 50 Hz in North America or Europe. see Fig. and relatively inexpensive. The attendant lack of ‡ uctuating "reaction torque" makes it possible to build and operate extremely large 3-phase generators which would otherwise be damaged by such ‡ uctuating torques.5 Design Considerations An AC machine (generator or motor) requires a large number of poles (Npoles ) by equation 2. reliable.12b in textbook. E. time.. in a four-pole machine the two coils for each phase can be connected in series (or in parallel because coil voltages are identical). The squirrel-cage rotor or reasonable facsimile is a common method of construction for induction motors.15 for a torque-speed curve of an induction motor. Otherwise there is no slip.12 in textbook.e.. ROTATING MACHINES due to relative motion between the rotor winding and the rotating armature …eld. I.g.19 if it operates at a relatively low mechanical RPM . In a three-phase system the electrical frequency is the same as in a singlephase system. there is no ‡ uctuating double-frequency torque component. Refer to Fig. On the other hand induction generators are rarely found except recently in wind power applications for which they are somewhat suitable. Hence in general the mechanical displacement angle between sets is (2 =3) = (Npoles =2) : E. Consequently induction motors are durable. these sets are symmetrically displaced by 120 mechanical degrees. Note that the induction motor must operate at less sychronous speed to develop torque. i. which will be shown by ux phasor addition of rotating waves. The e¤ect of a balanced three-phase system is to produce a rotating ‡ wave which has only a forward component.11.4 Three-Phase Machines Phase refers to the electrical phase or delay between AC currents in multiple armature windings. 4. 2. Here the "windings" are solid aluminum bars cast in slots and shorted together by common cast aluminum end rings. respectively.g. electrical frequency is determined solely by the number of poles in a machine. Likewise the currents supplied to or voltages generated from 3-phase windings are symmetrically displaced by 120 electrical degrees.11. 4. which are contributing factors to their immense popularity. 4. Hence three-phase motors possess constant torque vs. Rotor windings must be . Induction motors are the most common and widely applied of all motors.. Note that a frequency change and electromechanical energy conversion occur here. Most of the world’ power systems (hence most synchronous generators) are 3s phase. in a two-pole machine.54 CHAPTER 2. A 3-phase system can be either a synchronous machine or an induction machine where there are three sets of armature windings per pole-pair.

An AC machine (generator or motor) requires a small number of poles (Npoles ) by equation 2.. All turns between two diametrically opposed commutator segments are e¤ectively in series. by means of copper segments and carbon brushes. Typically by design (brush positions.17 shows an elementary DC machine with a commutator..2. consider a large hydroelectric generator where the prime mover is water power.1 shows such a generator where mechanical rotation is at 37 RPM. E. the stator …eld is inherently …xed whereas the rotor …eld is stationary with respect to the active positive and negative commutator segments.e.. respectively.e. so it is possible to deduce the output winding voltage as a function of time from the B …eld distribution.19 if it operates at a relatively high mechanical RPM. It turns out that the ideal stator B …eld is piecewise constant (a square wave) to maximize the DC component of the commutated (recti…ed) speed voltage. I. In a conventional DC machine this recti…cation is attained by placing a commutator and hence the armature on the rotor shaft while the (DC-excited) …eld winding is on the stator. This allows the rotor …eld to revolve in space by only the arc subtended by the active commutator segments. 4. In this case Npoles = 2 or Npoles = 1. 4. Round rotors (hence a uniform gap) are a necessity in this case because salient poles would incur unacceptably high stress and windage loss. s as shown previously.g. etc.11 shows such a generator. 4. The instantaneous induced speed voltage in a single turn of the rotor winding is proportional to the gap B …eld at the turn’ particular angular position . Fig.g.12. which is small if segments are numerous. Fig.g.. I. . Such speed voltage has zero average (DC) value so a form of recti…cation is required to achieve a non-zero DC value. in a salient-pole motor more space is available to accomodate rotor windings. Fig. The commutator mechanically "switches" the connection of the armature coils to external terminals in synchronism with machine rotation. Rotor windings are distributed so the B …eld will be determined only by the distribution and not by the gap. DC MACHINES 55 concentrated so the B …eld will be determined only by the rotor (gap) geometry and not by a winding distribution. E.) the stator and rotor …elds are perpendicular to each other to maximize torque. consider a large generator where the prime mover is a gas or steam turbine operating at 1800 or 3600 RPM. E. all rotating machines execute cyclic mechanical motion which creates a cyclic (AC) induced speed voltage in the windings to support electromechanical energy conversion. The spatial orientation between the rotor and stator …elds is nearly constant during machine operation.12 DC Machines A DC machine may be viewed as a "commutated" version of a synchronous AC machine. 2..

5 for the non-salient (uniform gap) and salient (non-uniform gap) cases.56 CHAPTER 2. to produce a prescribed driving mmf func- Only the …rst option is feasible with respect to salient poles where the windings must be concentrated.25) J( ) (2.21 reduces to 1 0 [B ( ) g ( ) B ( + ) g ( + )] = F ( ) (2. respectively. The …eld B ( ) varies periodically with due to continuous rotation around N the gap. Now we assume anti-symmetry of the current or current-sheet distribution and symmetry of the gap.21) C with F( )= Z J da (2. Equation 2. Variation of the gap length g ( ) versus 2. Note that the magnetic path C shown in each …gure passes through the origin and is symmetric with respect to it. This analysis is illustrated in Fig. round rotor AC machine). This implies anti-symmetry of the …eld: B ( + ) = B ( ): (2.13. where a rotor winding is shown for de…niteness.g. respectively.22) S where J is the current density through surface S. Only the second option is feasible with respect to non-salient poles where the gap must be uniform (e. the reference direction for B is everywhere radially outward. It will be shown below by magnetic-circuit analysis that two options are available to design or "shape" B ( ) : 1. by pole-face geometry. ROTATING MACHINES 2.. which are written as J( + )= and g( + ) = g( ). 4 and Fig.26) (2. and negligible reluctance is assumed for the stator and rotor core material. Application of Ampere’ law to path C yields s I H dl = F ( ) (2. Winding distribution versus tion F ( ). where the spatial electrical frequency is poles electrical cycles per 2 mechanical revolution.24) . The following magnetic-circuit analysis is applicable to B ( ) which is produced by either a rotor or stator winding.13 2.23) where B = 0 H in the gap.1 Design of Gap Field Magnetic Circuit Analysis Assume that a certain gap …eld B ( ) is required in the design of a given machine.

I.2 Idealized Current Density The idealized current density J which corresponds to a prescribed F ( ) follows from equation 2. which is converse to the process described above to …nd the idealized J ( ) from a prescribed F ( ) : However the idealized J ( ) still serves as a guide for modi…cation of a winding distribution to improve F ( ) : F ( ) is called the spatial mmf wave which so far is stationary in space. gives 1 0 57 [2B ( ) g ( )] = F ( ) 0F ( ) = 2g ( ) 0 Fag1 (2.19..22 when the linear segment of path C rotates through = 2 : Remark 6 The linear presentation of mmf Fag1 ( ) and gap g ( ) in Fig. 4 or Fig.13.25 and 2. 2. 4. Here Fag1 ( ) is simply a piecewise constant function (square wave) with values N i as shown in Fig. an abrupt change 2 of polarity occurs in the evaluation of equation 2.29) 2 is the mmf across one gap. these circular functions have been mapped to linear form or "laid out ‡at".23. by assuming that J is a thin current sheet on the surface of the rotor..13.e.19 is called developed form. J ( ) is the objective to be approximated by the turns distribution.3 MMF of Discrete Windings We determine the mmf F ( ) corresponding to certain discrete winding distributions by direct evaluation of equation 2. I.28) 2.28 shows that a required B …eld shape can be obtained by varying g ( ) (…rst option) or by varying F ( ) (second option). respectively. Fag1 ( ) = ( ) g( ) : (2. with subsitution of equations 2. Example 5 Consider the simplest case where an idealized concentrated rotor winding lies totally within a very small arc (approaching zero) located at = 2 : The rotor may be non-salient or salient.13.. where in either case the other variable is (at least piecewise) constant.26.27) or B( )= where F( ) (2. E. 4.g. Assume that this winding consists of N turns carrying the current i through the surface S.22 by di¤erentiation of both sides with respect to : In order to di¤erentiate we render the integral on the right-hand side of equation 2.e.22.2. 5. if F ( ) = sin then J ( ) = cos : In the case of a discrete winding. as illustrated in Fig. DESIGN OF GAP FIELD Hence equation 2. Equation 2. .22 as a function of (without r) in polar coordinates.

where the number of turns per slot (and possibly the spacing between slots) is varied. The actual windings are discrete with 24 total slots. Fag1 ( ) is a trapezoidal wave with c extreme values N2 i . Hence the linear transition in Fag1 ( ) is now approximated by 4 steps. In Example 5 the spatial wave Fag1 ( ) is a square wave with Fourier series representation "1 # X Ni Fag1 ( ) = An cos (n ) (2.28 the desired mmf wave Fag1 ( ) is a pure sinusoid if the machine has a uniform gap g ( ). which corresponds to a simpli…ed two-pole three-phase machine. n even 4 n. ROTATING MACHINES Example 7 Consider the armature winding shown in Fig. A linear transition between these extremes occurs with …nite slope due to the evaluation of equation 2.21.22 when the linear segment of path C is rotating through the current sheet J.10. 2.30) 2 n=0 | {z } unit square wave where An = odd 0. Refer to Example 4. Note that this analysis can be generalized to arbitrary Npoles simply by substitution of e for : 2. This is done to obtain an mmf wave Fag1 ( ) which is a closer aproximation to a sinusoid. where each side of the phase a winding occupies 4 slots.14 Gap Field in AC Machines In an AC machine the optimal gap …eld B ( ) is a pure sinusoid as explained in Section 2.21b. So the standard analytic approach is to represent the developed form of the mmf wave Fag1 ( ) by a Fourier series in the variable .1a.58 CHAPTER 2.4 Fourier Analysis of MMF Waves The spatial mmf wave Fag1 ( ) is periodic in . Now the distributed windings in some previous examples are assessed with respect to this objective. 4. n : (2. where the air gap is uniform (ignoring slots).13. 4. with period 2 for a two-pole machine. Hence by equation 2. Idealizing the phase a winding as a uniform current sheet J over one third of the circumference. as shown in Fig. 4.31) . Each phase winding consists of Nc turns uniformly distributed over one third of the stator circumference and wound in two layers to simplify access to end conductors. as shown in Fig. Fag1 ( ) approaches a pure sinusoid only if the Fourier coe¢ cients of all harmonics approach zero. Example 8 Consider the rotor winding shown in Fig. 4.b for an extension of this problem.20.20b. so our objective is that only the fundamental coe¢ cient be non-zero.

32.32) where b (Note that Fag1 ( ) is denoted in the textbook by simply Fag1 ( ) :) In Example 7 we consider a di¤erent spatial wave and denote it here by 0 Fag1 ( ) : Its Fourier series representation is 0 Fag1 ( ) = b max Fag1 ( ) = Ni 2 4 : (2. because Fag1 ( ) resides at its maximum value (‡ top) for a greater interval in : This leads at naturally to the de…nition of the winding factor kw = A0 1 A1 (2. i. (2.37) for any winding. A0 An n < A0 A1 1 for odd n sinusoid.. based on the condition that N i is the same as in the concentrated reference winding whose fundamental coe¢ cient is A1 .38) analogous to equation 2. The typical range of kw is 0:85 to 0:95: It is important to note that the winding in Example 7 does reduce the strength of harmonics relative to the fundamental. GAP FIELD IN AC MACHINES From the above series the fundamental component is b Fag1 ( ) = Ni 2 4 cos 59 (2.e.36) This can be shown from the formula for Fourier coe¢ cients.34) A0 6= An : n (2.2.39) 3: This follows because that winding is a closer approximation to a . (2.33) Ni 2 where in general 0 Fag1 3 1 X 7 6 A0 cos (n )5 4 n 2 n=0 n o dd (2. The fundamental component here becomes b0 Fag1 ( ) = kw N i 2 4 cos .35) The waveshape of ( ) is a closer approximation to a sinusoid but its funb damental component is actually smaller than that of Fag1 ( ) : A0 < A1 : 1 (2.14.

38 becomes b0 Fag1 ( ) = kw Nphase Npoles ia 4 cos Npoles 2 a . 4.22 and 4. This is equivalent to equation (4:5) in the textbook. for each phase equation 2. This is shown for two-pole and four-pole machines in Figs. 4. the peak value of the sawtooth mmf wave is max (Fag1 ( )) = where : Ca = total number of armature-winding conductors : m = number of parallel paths through armature winding ia = armature current.d and Example 4. N 2.40) where Nphase is the number of turns per phase (replacing N ) and ia is the 2 poles current per phase. 4. 4. respectively. In this last regard observe that the four-pole machine is more e¤ective than the two-pole machine in achieving the largest arc of gap uniformity. Hence by Fourier analysis of the sawtooth mmf wave. respectively. A. the peak value of its Ca 2mNpoles ia A turns/pole (2. ROTATING MACHINES In a 3-phase system.2. Elementary.17. The mmf Fag1 ( ) for these machines is shown in developed form in Fig.23 and Fig. and four-pole DC machines with concentrated (stator) …eld windings are shown in Figs.10. It is apparent from these …gures that the stator must be technically classi…ed as salient-pole even though the objective is a uniform gap over the largest possible arc with respect to the round rotor. In a DC machine the rotor winding must be distributed uniformly due to mechanical and operational constraints imposed by the commutator. By magnetic circuit analysis. two-pole. Refer to Example 4. where the spatial frequency of the sawtooth in the fourpole machine is twice that of the two-pole machine.15 Gap Field in DC Machines In a DC machine the optimal gap …eld distribution B ( ) is a (piecewise constant) square wave as explained in Section 2.24b.24.18. Hence by equation 2.1c. (2.24.60 CHAPTER 2. 4.22. respectively. 4.41) . For both machines the driving rotor mmf Fag1 ( ) is a stepwise approximation to a sawtooth wave. where the latter uses a current-sheet idealization for the rotor winding. The ideal uniform gap is only approximated over a large arc in a real machine due to physical winding and magnetic circuit constraints. which corresponds to a concentrated (stator) …eld winding if ideally the machine has a uniform gap g ( ).28 the desired mmf wave Fag1 ( ) is also a square wave. The actual "‡ at-topped" B …eld obtained in an elementary two-pole machine is shown in Fig. and 4.

. In complex cases a general …eld solution is necessary. GAP FIELD FOR NON-UNIFORM AIR GAP spatial fundamental component is b max Fag1 ( ) = 8 2 61 Ca 2mNpoles ia : (2.28 may remain valid in the regions where the gap is small (and usually piecewise uniform). .15.2.42) 2.26b.17 2. These extreme cases fall by de…nition into the category of salient-pole machines. …nite-element analysis). whereas machines under previous consideration possessed air gaps that were uniform or a least piecewise uniform over a large arc. In a balanced three-phase system. The analysis which culminated in equation 2. A corresponding decomposition of each phase in a three-phase system is possible. This is essentially a reproduction of Fig. Sometimes an analytic solution is still possible in such cases by development of other simplifying assumptions (later).27.g.16 Gap Field for Non-Uniform Air Gap The variation of air-gap length g ( ) is extreme in certain machine designs. However that analysis does not apply in the region of this machine where the gap is very large because its simplifying assumptions are usually invalid there. An example is the salient-pole DC machine shown in Fig. 4.1 Rotating MMF Waves (AC Machines) Introduction For a single AC phase.26.22 which was previously discussed in Section 2. This can be achieved by use of numerical methods (e.16. Note that the analysis here is also applicable to the harmonics of each phase if necessary. A decomposition of this wave into the sum of a forward (travelling) wave and a backward (travelling) wave is possible. the superposition of all forward waves reinforces to produce a large net forward wave whereas the superposition of all backward waves cancels to produce a zero net backward wave. 4.17. consider the fundamental component of the spatial MMF wave which is stationary in space while it undergoes a "pulsating" amplitude variation with respect to time. A …nite-element solution for the magnetic …eld in a salient-pole DC generator is shown in Fig. 4. Likewise that analysis does not apply in a machine where the gap is large everywhere. 2. 4. These reinforcement and cancellation properties are shown by phasor analysis to be the result of phase di¤erences or time delay between the individual-phase MMFs. which yields a forward wave and a backward wave for each phase. An example of this is the salient-pole AC synchronous machine shown in Fig.

43) ae ] [cos ! e t] (2.50) Fmax 2 [cos ( ae ! e t)] (2.46) b b It is understood that Fag1 ( ) is really Fag1 ( .2 Decomposition of Single Phase For a single phase winding.45) and : Fmax = kw Nphase Npoles Ia : (2. t) in equation 2.44 where we take = and = ! e t: This yields where the forward-travelling wave is + : Fa = + Fa = Fa + Fa ae (2.28.62 CHAPTER 2.44 but the latter b notation is omitted for simplicity. this gives where b Fag1 ( ) = Fmax [cos ae (2.travelling waves.17. ROTATING MACHINES 2. b The decomposition of Fag1 ( ) into the sum of forward and backward travelling waves is accomplished simply by applying the trig identity cos cos = 1 2 [cos ( ) + cos ( + )] (2.28c.49) (2.52) Here the subscript a refers to phase a and the superscripts + and refer to forward. .and backward. 4. refer to equation 2. 4. This decomposition is illustrated in phasor form by Fig. Note that here Fag1 ( ) remains …xed in space with the envelope cos ae while it amplitude is pulsating with respect to time due to the factor cos ! e t: Refer to Fig. respectively.51) and backward-travelling wave is : Fa = Fmax 2 [cos ( ae + ! e t)] : (2.48) (2.40 for its fundamental spatial mmf wave (which is stationary in space) and assume the sinusoidal excitation ia = Ia cos ! e t: b0 Dropping the prime notation in Fag1 ( ) .47) to equation 2.44) : Npoles = 2 4 a (2. Note that the subscript ag1 (referring to "air gap 1") and the decoration b (referring to fundamental spatial component) have been dropped to simplify notation but are understood.

2.17. ROTATING MMF WAVES (AC MACHINES)

63

2.17.3

Decomposition of Three-Phase System

Here we simply apply the above result for a single-phase to each phase of a balanced three-phase system; refer to Fig. 4.29. This requires only modi…cation of the angle in equation 2.43 to indicate that the currents of phases a; b; and c are symmetrically displaced in phase by 120 : Hence in phasor notation
+ Ia = Im \0 =) Fa = Fa + Fa ;

(2.53) (2.54) (2.55)

Ib = Im \ and

+ 120 =) Fb = Fb + Fb ;

+ Ic = Im \120 =) Fc = Fc + Fc :

The net mmf is one wave F = Fa + Fb + Fc which can be decomposed into forward- and backward- travelling waves: F = F+ + F : Clearly
+ + + F + = Fa + Fb + Fc

(2.56)

(2.57)

(2.58) (2.59)

and F
+ : Fa = + : Fb =

= Fa + Fb + Fc : Fmax 2

The individual forward-wave components are [cos (
ae

! e t)] ; Fmax 2

(2.60)

Fmax 2

[cos (

ae

120

[! e t

120 ])] =

[cos (

ae

! e t)] ; (2.61)

and
+ : Fc =

Fmax 2

[cos (

ae

240

[! e t

240 ])] =

Fmax 2

[cos (

ae

! e t)] ;

(2.62) where the cosine arguments ae ; ae 120 ; and ae 240 are due to the spatial o¤sets of windings a; b; and c; respectively, and the cosine arguments ! e t; ! e t 120 ; and ! e t 240 are due to the time (phase) o¤sets of ia ; ib ; ic ; respectively. Hence : F+ = from equation 2.58. 3Fmax 2 [cos (
ae

! e t)]

(2.63)

64

CHAPTER 2. ROTATING MACHINES Likewise, the individual backward-wave components are : Fa = Fmax 2 [cos (
ae

+ ! e t)] ; Fmax 2

(2.64)

: Fb = and Fc : = =

Fmax 2

[cos (

ae

120 + [! e t

120 ])] =

[cos (

ae

+ !e t

240 )] ;

(2.65) Fmax 2 Fmax 2 Fmax 2

[cos ( [cos (

ae

240 + [! e t + !e t 120 )]

240 ])] =

[cos (

ae

+ !e t

480 )] (2.66)

ae

F will be found from equation 2.59 where the required summation will be simpli…ed by using phasor form. (Note that phasor form can only be applied to groups of waves travelling in the same direction because a reversal of direction is equivalent to reversal of signs between spatial and temporal arguments in the cosine function.) The phasor assignments Ia ; Ib ; and Ic induce a phasor interpretation on Fmax by means of equation 2.46 interpreted for each phase current. + + + In turn this induces phasor interpretations on Fa ; Fb ; Fc and Fa ; Fb ; Fc through equations 2.51 and 2.52, interpreted for Fmax of the respective phases. Hence we can write Fmax e Fa = \0 ; (2.67) 2 and e Fb = Fmax 2 \120 (2.68)

where the decorationeindicates the phasor representation of a variable. Applying equation 2.59 in phasor form: e F = Fmax 2 [1\0 + 1\120 + 1\ 120 ] = 0: (2.70)

e Fc =

Fmax 2

\

120 :

(2.69)

Hence we conclude that F = F + + F = F + consists of only a forwardtravelling spatial mmf wave whose amplitude is three times that of the forwardtravelling wave due to a single phase acting alone. Theorem 9 The forward-travelling wave rotates at the synchronous angular velocity 2 : !s = !e : (2.71) Npoles

2.18. ANALYSIS OF NONSALIENT-POLE MACHINES Proof. At a given point on the advancing wave
ae

65

! e t = constant ! e t = constant.

or

Npoles 2

a

For a time interval 4t this gives Npoles 4 2 or
a

! e 4t = 0

This last expression is the actual rate of angular advance and agrees with ! s : Refer to Examples 4.3 and 4.5. Refer to Practice Problem 4.4.

4 a !e i: =h Npoles 4t
2

2.17.4

Machine Analysis by Flux Waves

The decompositions in the preceding sections can be applied directly to the analysis of two-pole single-phase and two-pole three-phase synchronous motors rotating at synchronous speed. The forward mmf wave F + produces a B …eld rotating forward which pulls the rotor in synchronism with it, hence producing a constant forward torque on the rotor winding which carries DC current. The reverse mmf wave F produces a B …eld rotating backward at 2! m with respect to the rotor hence a double-frequency reverse torque on the rotor winding with average value zero. The total torque is the superposition of the forward and reverse torques. In the single-phase case the torque is net forward but carries the undesirable double-frequency variation. In the three-phase case the backward torque is zero because the backward-travelling mmf wave is zero. So in the three-phase case the net forward torque is absolutely constant. General AC synchronous machine operation can be characterized in terms of ‡ waves as follows. In a generator the rotating …eld of the rotor pulls the ux forward-travelling armature ‡ wave (due to mmf wave) along at synchronous ux speed while the electromechanical torque opposes rotation which is sustained by the prime mover. In a motor the rotating forward-travelling armature ‡ ux wave (due to mmf wave) pulls the rotor along at synchronous speed while the electromechanical torque aids rotation.

2.18
2.18.1

Analysis of NonSalient-Pole Machines
Introduction

We analyze an idealized elementary nonsalient-pole (uniform air gap) machine with distributed stator and rotor windings on cores with negligible reluctance.

. and d dt is the time-derivative operator. Refer to Appendix B. ROTATING MACHINES This analysis can be easily generalized later to more complex machines. The stator and rotor ‡ linkages are given by: ux s r = Lss Lsr ( me ) Lsr ( me ) Lrr is ir : (2.73) where the maximum Lsr occurs when the magnetic axes of the stator and rotor are aligned ( me = 0).75 gives vs vr = Rs 0 0 Rr is ir + d dt Lss Lsr ( me ) Lsr ( me ) Lrr dis d + [(Lsr ( dt dt is ir (2. ux 2. and rotor-to-stator mutual inductance are denoted by Lss . 4. rotor self inductance.66 CHAPTER 2.3 Machine Inductance Parameters The stator self inductance.75) where Rs and Rr are the stator and rotor winding resistances. It is shown there that Lsr ( me ) = Lsr cos ( me ) (2.74 into equation 2.34 which shows a machine for the special case Npoles = 2: Optimal b operation is assumed.18. respectively. (2.74) 2. the spatial mmf wave Fag1 ( ) hence the spatial magnetic ‡ wave B ( ) are pure (fundamental frequency) sinusoids.76) or (in expanded form) vs = Rs is + Lss me )) ir ] (2.77) .4 Network Equations vs vr Rs 0 0 Rr is ir d dt The terminal voltages are given by: = + s r (2. Substitution of equation 2. Refer to Fig.18. 2. The machine is treated as an electromechanical system with two electrical terminalpairs and one mechanical terminal-pair. i. Lrr . respectively.72) where m is the angle between the axes of the stator and rotor windings.18. where the italic capital letters denote constant values and the script letter denotes a variable.e. and Lsr ( me ) .3 (textbook) for the derivation of these inductances with respect to an assumed spatial fundamental (sinusoidal) mmf wave.2 Machine Speci…cation Npoles 2 The derivation applies to a general multipole machine where me = m.

All self inductances will be constant and mutual inductances will be constant amongst all windings on one side of the gap (stator or rotor).80 into equations 2.79 and 2.70) in the textbook. Examination of the mutual interaction voltages shows vr = Rr ir + Lrr d [(Lsr ( dt me )) ir ] = d (Lsr ( dt d (Lsr ( dt me )) ir + [Lsr ( me )] dir dt dis : dt (2.80) Note that in each of these last two equations the terms (from left-to-right) are speed voltage and mutually induced voltage (transformer action). analogous terms in the case where there are multiple stator and/or multiple rotor windings. respectively. m) Lss i2 Lrr i2 Npoles s r + +Lsr is ir cos 2 2 2 (2.5 Coenergy by Inductance Parameters This is a multiply-excited magnetic-…eld system which was previously analyzed for the general case.78) dt dt Note that in each of these last two equations the terms (from left-to-right) are resistive (loss). self-induced voltage. = me )] is ir Lss i2 Lrr i2 s r + +[Lsr ( 2 2 = m : .78.81 into equations 2. L22 ( ) = Lrr .80 followed by substitution of equations 2. and mutual interaction voltage.81) me )] dir dt [Lsr ir sin ( me )] me dt (2.77 and 2. the identi…cation L11 ( ) = Lss . Only mutual inductances between windings on opposite sides of the gap will be functions of m : Note that methods here can be applied in cases where core permeability is …nite.83) dt dt dt by substitution of equation 2. With reference to that equation. Relevant to the speed-voltage term we evaluate: d d (Lsr ( me )) = (Lsr cos ( dt dt Finally we have vs = Rs is + Lss and dis d me dir + [Lsr ( me )] [Lsr is sin ( me )] (2.79) and d [(Lsr ( dt me )) is ] = me )) is + [Lsr ( me )] (2. Refer to equation 1.18. vr = Rr ir + Lrr dis + [Lsr ( dt me )) = [Lsr sin ( me )] d me dt d : (2. and L12 ( ) = Lsr ( me ) yields the coenergy 0 Wf ld (is .82) 2.84) This expression for coenergy generalizes to include additional.72 in the previous notes entitled "Electromechanical Energy Conversion" or equivalently equation (3.18. ir .2. : me Note that ! me = d dt . ANALYSIS OF NONSALIENT-POLE MACHINES and 67 dir d + [(Lsr ( me )) is ] : (2.79 and 2.

These stator and rotor mmf waves must have the same spatial frequency but there is no restriction with respect to their relative motion. into the previous equation gives i2 h 0 0 Fag1 ( ) 2 @Wf ld g Fag1 ( ) 0 = = . ROTATING MACHINES 2.6 Coenergy by Magnetic Field The coenergy is found directly here by integration of the average magnetic-…eld coenergy density over the volume of the air gap.35 which shows Fsr as the diagonal of a parallelogram with sides Fs and Fr and included angle sr : Hence by trigonometric formula 2 2 2 Fsr = Fs + Fr + 2Fs Fr cos sr : (2. Refer to Fig. toothtip. end-turn.18.e. The peak magnitudes of the stator and rotor mmf waves are denoted by Fs and Fr .89) 4 Fsr g 2 . B and H are in the gap and functions of . Such leakage ‡ can be categorized as stator leakage ux ux or rotor leakage including slot.88) where Fag1 ( ) is the sinusoidal mmf wave which is the sum of the stator and rotor mmf waves.. and we have made the substitution H = B . Previously we have denoted the peak magnitude of this wave by Fsr : For integration over the gap volume it is su¢ cient to use the average value of @Wf ld @V over rotation around the gap from = 0 to = 2 : ( ) ( ) 0 2 @Wf ld Fag1 ( ) 0 Average = Average @V 2 g oi h n 2 0 = Average [Fag1 ( )] 2g 2 = 0 2g 2 2 Fsr = 2 0 0 (2.68 CHAPTER 2. We assume that the magnetic ‡ density B ( ) in the gap is radial and due to the sum (superposition) of the ux stator and rotor spatial fundamental (sinusoidal) mmf waves. .87) (2. @V 2 0 2 g B= 0 Fag1 ( ) (2. Only the magnetic ‡ which links both stator and rotor coils ux contributes to the mutual coenergy in the gap so we are in e¤ect assuming that leakage ‡ is negligible. respectively. while the spatial angle between these waves is denoted by sr : The peak magnitude of the resultant mmf wave (sum of stator and rotor mmf waves) is denoted by Fsr : Fsr is the phasor sum of Fs and Fr where it is su¢ cient to use only the angle sr for the present analysis. Substitution of 0 The coenergy density in the gap is 0 h i h i @Wf ld = 0 H2 = 0 @V 2 2 B 0 2 = (2. 4.85) B2 . 2 0 where V denotes volume. i.86) . either wave may be moving (rotating) or stationary.28. and space-harmonic leakage. g which follows from equation 2.

92 to the coenergy result in equation 2.85. This gives T = Npoles Lsr is ir sin 2 Npoles 2 m . we apply equation 2.94) . so V = ( Dl) g.i2 = ir .92) is =constant ir =constant which is equation 1.90) Wf ld = Average @V 4 g where l and D are the axial length and mean diameter of the air gap. integration over the gap volume: " )# # " ( 0 2 @Wf ld 0 Fsr 0 V = [( Dl) g] . we apply equation 2. (2.92 to the coenergy result in equation 2. (2.7.93) where the negative sign indicates that the torque T acts to decrease m hence align the axes of the stator and rotor windings. Refer to Examples 4.2. Finally.73 in the aforementioned notes.91. Note that methods here can be applied in cases where core permeability is …nite. the above torque equation would include additional. 2. The torques would still act to align the magnetic axes of corresponding windings. In the case of multiple stator and/or multiple rotor windings.84.7 Torque from Coenergy # In general the torque is given by " 0 @Wf ld (is . and = m have been made: The torque will be determined by applying this equation to the two di¤erent coenergy expressions which resulted by the inductance method and the magnetic-…eld method. respectively.18. where the identi…cations i1 = is . This gives T = 0 Dl 2g Fs Fr sin sr (2. analogous terms corresponding to all mutual inductances "Lsr " between windings on opposite sides of the gap.6 and 4.84.18. as described in the comments following equation 2. Regarding the inductance method. This can be rewritten as Wf ld = 0 0 Dl 4g 2 Fsr = 0 Dl 4g 2 2 Fs + Fr + 2Fs Fr cos sr (2. Regarding the magnetic-…eld method. ANALYSIS OF NONSALIENT-POLE MACHINES 69 where the last line follows because the average of the square of a sinusoid is one-half of the square of its peak value. ir .91) where the …nal expression follows by substitution of equation 2. T = @ m m) (2.

It follows that positive and negative values of sr correspond to generator and motor action. 2 2 Note that the terms Fs and Fr contributed nothing to the partial derivative because the condition that is and ir be constant is equivalent to constant Fs and Fr . increased torque requires increased area ( Dl) of the gap (hence increased surface area of the rotor and stator).98 can be rewritten as T = Npoles 2 2 sr Fr 2 sin r. Equivalence of the torque equations 2. (2. (2. Analogous to the result by the inductance method.. which generalizes to T = Npoles 2 0 Dl 2g Fs Fr sin sr (2.95) for the multipole case. Bsr is limited to about 1:5 to 2:0 T by magnetic saturation of armature teeth and Fr is limited by constraints on rotor winding current related to temperature rise. where an orthogonality condition is operative between the mmf waves of di¤erent poles during the integration of magnetic-…eld energy density.98) which follows from equation 2.99) .70 CHAPTER 2. respectively. ROTATING MACHINES for the assumed two-pole case. by equation 2.95 can be interpreted as the product of quadrature components of the stator and rotor mmf waves.e.97) where the angles s and r are taken between components corresponding to stator and resultant and corresponding to rotor and resultant. respectively. etc. Hence given speci…c limits on Bsr and Fr while Npoles is …xed. (2. the negative sign indicates that the torque T acts to decrease sr hence align the axes of the stator and rotor windings. as shown in Fig.95 can be shown by invoking results for Lsr from Appendix B. 4. Equation 2. by associating the term sin sr with either Fs or Fr in this product.3 (textbook) and previous magnetic circuit analysis for Fs and Fr : The torque given by equation 2.93 and 2. This generalization can be obtained by superposition of all mmf waves.35.98.97 by substitution of Fsr = gBsr : Here resultant 0 ‡ density Bsr and mmf Fr appear explicity so this idealized result can be ux used to determine maximum torque for a given size of machine.96) 0 Dl 2g Fr Fsr sin r. I. Another useful expression for torque is T = Npoles 2 Dl 2 Bsr Fr sin r. Alternative expressions (for application in appropriate circumstances) are T = and T = Npoles 2 Npoles 2 0 Dl 2g Fs Fsr sin s (2.

ir .103) or Bsr = Dl : (2. and me . we have ux sr = [average fjBjg] Ap (2. and instantaneous values of Fs .1 System Analysis Elementary Machine Model The elementary non-salient pole machine is by assumption a lossless magnetic energy storage system or device in the preceding analysis. 2. sr .19. the average fjBjg is taken over one-half wavelength. such device analysis was originally introduced in the notes entitled Electromechanical Energy Conversion. SYSTEM ANALYSIS where sr 71 is the total ‡ resultant over the area of one pole.102) Dl Npoles (2. I.. The "across" variabes are vs and vr at the electrical ports and T (torque) at the mechanical port. while at each port there is one "through" variable and one "across" variable.98 gives the result in equation 2.93 specify the relationships between the variables at the terminal-pairs (ports) of the system. and : Ap = is the surface area of one pole. there are two electrical ports and one mechanical port. and 2. In each of the three system equations a di¤erent "across" variable is a function of all the "through" variables.82.19 2. Note that these system equations and the alternative expression for torque in equation 2.6. Refer to Example 4. Hence these equations are in general applicable to any type of machine. ir .95 are expressed in terms of instantaneous values (and time derivatives) of is .100) where B is the magnetic …eld (‡ density) resultant from combined stator and ux rotor mmfs. or even DC. The only underlying assumption in their derivation was a uniform-gap machine.83. 2.19. whether it be AC synchronous.8 and Practice Problem 4. Now average fjBjg = 2 Bsr (2.e. . In this elementary machine.101) where Bsr is the peak value of the sinusoid B. so sr = 2 Bsr Dl 2Dl = Bsr Npoles Npoles Npoles 2 sr (2. AC induction. Fr . respectively.99. Here the resulting system equations 2.2. me .104) Finally substitution of this last expression into equation 2. respectively. The corresponding "through" variables are is .

and an external torque Text on the rotor moment of inertia. However in some cases these system equations can be specialized (e. For a generator we may take the applied stator voltage source as zero and use vs now to denote the voltage across the load Req : Text plays the role of either an opposing load torque or an applied primemover torque according to the intended operation of the machine as either a motor or a generator. The additional elements are: a resistance Req in series with or in place of the applied stator voltage source vs .93. In either case. . respectively.106) For a motor Text is a negative function of speed because it opposes rotor rotame tion.105.2 Practical Machine Model For a basic practical machine model we assume that the rotor moment of inertia JR and the rotor and stator winding resistances Rr and Rs are added external to the elementary model. In either case. it can be incorporated into the system equations simply by the replacement (Req + Rs ) ! Rs in equation 2.83.. In either case. a torque-speed characteristic encompasses many e¤ects within a prime mover including viscous friction associated with ‡ uid (e.82. e.72 CHAPTER 2.g. Now this model is generalized slightly to cover the common cases of motors and generators by adding two more elements.82 and 2. with viscous friction Text = kf d dt where kf is the coe¢ cient of friction.. where additional constraints between port variables can be imposed by externally connected electrical and mechanical elements. 2. it can be incorporated into the system equations simply by the replacement (Text + T ) ! T in equation 2. Req plays the role of either a source or a load resistance according to the intended operation of the machine as either a motor or a generator.g. The winding resistances were included already in the electrical-port equations 2. ROTATING MACHINES These nonlinear di¤erential equations govern system behavior.105) by application of equation 2. fuel) ‡ as ow well as other mechanical motion.. the functions for Text can be represented by a Taylor series if increased accuracy is required.g. Complete analytical or closed-form solutions of these equations are generally not possible so numerical methods are appropriate. which gives = d! me Npoles = dt 2 Text + T JR = Npoles 2JR Npoles [Lsr is ir ] sin ( 2 me ) Npoles d! me = dt 2 d dt Npoles 2 T = JR Npoles 2 2 Text : (2. The moment of inertia gives Lsr is ir sin ( me ) JR (2. For a generator the applied Text due to a prime mover may be taken as a positive constant over a limited range of speed but more precisely it is me a function of speed ( d dt ) given by the prime-mover torque-speed characteristic.19. respectively. In general. linearized about an operating point) to expedite a local analytic solution.

u(t).e. hence their magnetic …elds. t) (2.19.. and me are state variables of the elementary . The "evolution" of state variables (second property) is expressed mathematically by the general (nonlinear and time-varying) state equations in vector form: dx = f (x(t).2.108 represents a set of output equations where g (x(t). An output is a variable which is a function of state variables and input but otherwise completely unconstrained.19. scalable respectively by ir and is . given present and future inputs. future values of the set follow from the present values of the set.107) dt and y (t) = g (x(t). time-varying function. u(t). SYSTEM ANALYSIS 73 2. the set determines all other system variables. Equation 2. is . dependency of all other variables on state variables) must hold while minimality implies no direct functional dependencies (i.108) where x(t). t) . are: …xed geometrically with respect to the individual rotor and stator axes. Note that state equations can be simpli…ed by using matrix form if the system is linear but electromechanical systems generally involve nonlinear terms. an ideal voltage source. u(t). at an arbitrary instant in time. These current distributions. inputs.. time-varying function. 2. state variables are de…ned as any minimal set (vector) of variables with the following properties: 1. respectively. and y (t) denote vectors of state variables.107 represents a set of …rst-order di¤erential equations where f (x(t). An input is a completely constrained or speci…ed variable (function of time) "applied" somewhere in the system.19. The second property indicates that knowledge of past inputs is not necessary to determine present and future system behavior because the present values of state variables summarize the past or "history" of the system. which is fundamental to system theory. implies independence) amongst the state variables themselves. Input and output roles must be interpreted or assigned within the context of a given system.4 State Variables for Machine Models The elementary machine model is an idealized lossless magnetic energy storage system characterized by its rotor and stator spatial current distributions. The …rst property (i. u(t). Equation 2.g.. t) can be a nonlinear. It can be shown that ir . For a given system. and movable only with respect to each other by rotation based on the parameter me . is introduced brie‡ here as a conceptual aid and as a connection to more advanced topics y which are beyond the scope of this course. u(t). In addition it can be shown that the energy stored in a system is a direct function of the state variables. t) can be a nonlinear. and outputs. (2.e. e.3 State Variable Concept The state-variable concept. 2.

Only JR has any e¤ect with respect to required state variables. based on the intended machine operation as motor or generator. For s a motor. while the external torque Text is opposing and represents . x4 . However now the restoring force is provided by the magnetic …eld which acts like a "spring".83. where me is regarded as an arbitrary but …xed parameter as if the rotor were held in place by a mechanical clamp. vs . otherwise the magnetic torque T would cause d dt The practical machine model was produced from the elementary machine model by the addition of external elements: rotor moment of inertia JR .19. Here the state vector is 2 3 x1 6 x 7 x=6 2 7 4 x3 5 x4 : : : : where x1 = is . and x4 = ! me : We require expressions for dx4 dx1 dx2 dx3 dt . now one must introduce the additional state : me variable: ! me = d dt : Note that taking JR ! 1 would impose d!me = 0 dt me hence d dt = ! me = 0. This (temporary) constraint on d me me gives dt = 0. plus winding and load resistances.106.82. These "knowns" (or corresponding dt known expressions) play the role of coe¢ cients of the "unknowns" or the role of constants. : me there d dt = dx3 = x4 is known as well. me The state variables me and d dt are analogous to the position and velocity of a mass which undergoes translational motion where a restoring force is provided by a spring. and 2. dt . x3 .74 CHAPTER 2. x2 = ir . x3 = me . because it is a kinetic-energy storage element rather than a loss or source element. the other input is the applied stator (armature) voltage vs and the : output is normally taken as the angular position x3 = me and/or the angular : velocity x4 = ! me .. The expressions for dx1 = dis and dx2 = dir dt dt dt dt dt follow by solving equations 2. me ! 1.106 and dx3 = x4 . 2. where dt dt x1 . Note that the expression for dx4 is given already : dt : by equation 2. ROTATING MACHINES machine. It remains to assign inputs and outputs to the resulting state equations.83 for the "unknowns" dx1 and dx2 . 2. The actual solution is left as an exercise for the reader. so the method of solution is analogous to solving a 2-by-2 system (two equations and two unknowns) of linear equations.82 and 2. x4 . dt . the inputs. x2 . one (albeit …xed or slowly varying) input is the applied rotor (…eld) voltage vr because it serves only to establish the machine’ magnetic operating …eld.107 will follow from equations 2. vr and all machine parameters are assumed given or "known". necessary due to the assumption that the rotor is massless. and the machine parameters. in the same vein as the aforementioned mechanical clamp which is now removed.and dt as functions of the state variables x1 .5 State Equations State equations for the practical machine in the form of equation 2. There potential energy is a function of position and kinetic energy is a function of velocity. x2 . I. In any case. x3 . a torque on JR .e.

Otherwise the torque T on the massless rotor : would cause its instantaneous alignment with the stator axis such that x3 = me = 0 permanently. The desired criterion will follow simply by determining if the magnetic …eld is delivering or absorbing mechanical power. x3 = me and are basically a subset of the state : equations for the practical machine.. Note that the coenergy was found already in equation : : : 2. in which case the DC motor is acting as a generator. SYSTEM ANALYSIS 75 : a mechanical load (e.19.108 depends on the intended operation of the machine. the machine does not rotate backwards though it may decelerate occasionally. (2. E. irrespective of its intended mode of operation.84. as a function of the state variables x1 = is . Instantaneous mechanical power output (delivered) at the rotor shaft is p = T !m . x3 = me : State equations for the elementary or idealized machine require only the : : : state variables x1 = is .2.6 Motor and Generator Operation It remains to derive a simple criterion which indicates if a machine is operating as a motor or as a generator at any given time. x2 = ir . we assume ! m > 0.e... viscous friction) as a function of x3 = me : For a generator. respectively. motors may be operating as generators (temporarily) and vice versa.. The connection between motor operation (T > 0) or generator operation (T < 0) and the machine state is found by interpretation of equation 2. Conversely the rotor wave leads or "pulls" the stator (armature) wave in a generator. The resulting state equations are nonlinear but amenable to solution by numerical integration.19. the other input is the external applied torque Text due to a prime mover and the output is taken as the stator (armature) voltage vs across the load Req .95. This would give zero magnetic torque T and the system would function thereafter simply as two mutually coupled inductances. x2 = ir .g.109) where T is the instantaneous torque on the rotor produced by the magnetic …eld. an AC synchronous generator may be forced into momentary motor operation to "sync up" when connected to a power grid (so called "in…nite bus")..g. Clearly motor: T >0 sr < 0 if (2. i. in an electric vehicle regenerative braking with concurrent battery recharge can be accomplished by temporarily reducing the armature voltage of a DC motor to zero. 2. .110) generator: T < 0 sr > 0 since Fs > 0 and Fr > 0: Recall that sr is the instantaneous angle of the stator B-…eld ("‡ ux") wave relative to the rotor B-…eld ("‡ ux") wave. Hence the stator (armature) wave leads or "pulls" the rotor wave in a motor. I. Clearly power delivery (motor operation) and power absorption (generator operation) correspond to p > 0 (hence T > 0) and p < 0 (hence T < 0). Here the assumed constraint of x3 = me to a constant value is necessary.e. Hence the output form of equation 2.g. E.

76

CHAPTER 2. ROTATING MACHINES

The torque T , hence its sign, is also given by the fundamental equation 2.93 showing the connection to the angle m . However this equation is more di¢ cult to interpret for the present purpose because m is the angle between the rotor and stator magnetic axes which is not known explicitly. A more complete analysis of m would be redundant in that it is analogous to the process which was used to derive equation 2.95.

2.19.7

Application

Complete solution of the system di¤erential equations in their original form or in state-variable form is di¢ cult due to the inherent nonlinear terms. However in the important AC steady-state case the solution becomes less complex. So many problems related to speci…c types of machines are considered under AC steadystate conditions in the sequel. Complete solutions which include transients require more advanced (usually numerical) techniques.

.

turbine generators.2 Machine Construction Synchronous machine construction may utilize either a non-salient pole (cylindrical) rotor (see Figs.and mutual inductances.1 Introduction A detailed analysis of AC polyphase (actually three-phase) synchronous machines will be presented here. In particular. equation 1.10 and 4. However basic principles carry over intact from the elementary machine to the polyphase synchronous machine. torque. 3.3 Excitation DC excitation is applied to the …eld winding which is almost invariably on the rotor of the synchronous machine. The exciter (machine) is usually mounted on the same shaft as the rotor. air gap …eld. with regard to winding self. DC exciters (found in older machines) supply 79 . Salient pole rotors are more suitable for slow-speed multi-polar hydroelectric generators and for most synchronous motors. Some underlying principles of synchronous machines were given previously in Section 1. 4. The analysis here amounts to a generalization of the elementary non-salient pole machine presented in the aforementioned notes.9).Chapter 3 Synchronous Machines 3. Non-salient pole rotors are more suitable for high-speed gas. The polyphase synchronous machine includes three stator (armature) windings (one per phase) and one rotor winding whereas the elementary machine has one stator winding and one armature winding.18 is the connection between electrical and mechanical frequencies whereby the rotor rotates in synchronism with the magnetic …eld produced by the armature currents. 3.or steam.11 of the notes entitled Rotating Machines. etc.11) or a salient pole rotor (see Fig. energy and coenergy. 4.

The voltage and frequency at its armature terminals are constrained by the overwhelming presence of numerous other. AC exciters (found in newer machines) supply DC by means of recti…ers where (depending on the con…guration of the exciter) the supply connection is either through slip rings or rotating with the shaft (hence a "brushless excitation" system).e. for R and Ff constant. This problem is an area of active research in power engineering. SYNCHRONOUS MACHINES the …eld winding through slip rings because their DC output must be taken from brushes which are stationary with respect to the commutator (on the rotor of a DC machine).) The torque-angle characteristic of equation 3. Hence the individual machine has a component of armature current (magnetic-‡ wave) ux which rotates at the network synchronous frequency (fe ) and tends to "lock in" the machine’ mechanical frequency (fm ).1.80 CHAPTER 3. (Note that the textbook omits the negative sign with this understanding. The excitation power is typically only one to two percent of the synchronous machine rating. hundreds of) synchronous generators operating in parallel over thousands of miles of transmission lines.1 is plotted in Fig.1) : : where R = resultant air gap ‡ (per pole). and RF ! r : The torque T acts in a direction to bring the magnetic …elds into alignment. the res mainder of the network seen by the individual machine can be modeled as a constant-voltage. ux : RF = electrical phase angle between magnetic axes of R and Ff : This equation follows directly from equation 1. 2 (3.5 Torque 2 R Ff The torque produced by the magnetic …eld at the rotor shaft is T = Npoles 2 sin RF . I. in extreme cases this can lead to a widespread power outage. Likewise reliability should increase because backup power sources are available if a speci…c machine fails. 3. For analytic purposes. Ff = mmf of DC …eld winding. However a serious network drawback is possible loss of synchronism due to large transients initiated by a disturbance.99 in the aforementioned notes simply by the following identi…cations: R ! sr . 3. generating capacity can be shared by di¤erent regions which experience peak demand at di¤erent times.4 Network Operation The US power grid is a network containing many (e. synchronized machines over the network. 5. constant-frequency source ("in…nite bus").. Ff ! Fr . as indicated by the negative sign. An individual machine operates in a network as follows. Most electrical energy is produced and distributed this way because of economic considerations regarding plant investment and operating costs.. In this plot the regions corresponding to .g.

while like subscripts (as in Laa ) indicate a self-inductance and unlike subscripts (as in Lab ) indicate a mutual inductance. The script L denotes the possibility that an inductance may vary with rotor position me . cg and the other subscript is from the set ff g: e. so maximum jT j is called the pull-out torque. cg or if both subscripts are from the (trivial) set ff g: e. In this case both windings lie on the same side of the gap. The torque angle RF must increase if the machine torque T must increase. i.6 Inductance Parameters where the subscripts a. Lbf . then the machine "pulls out" of synchronism. Lab . b. Lf a .g. In the sequel. etc. step) increase in mechanical or electrical load will cause a "hunting transient". INDUCTANCE PARAMETERS 81 motor and to generator operation are identi…ed by means of the criterion in Section 1. This occurs for a motor if the opposing torque increases due to a greater mechanical load or for a generator if the applied prime mover torque must increase to supply a greater electrical load..74 which gives the ‡ ux linkages for the elementary machine in the aforementioned notes.6.e. Also the subscript r corresponding to the single rotor winding The ‡ linkages of the two-pole three-phase machine in Fig. etc.g.19. rotor side) of the gap. Lf f . 3. in view of equation 3. Analysis of the hunting transient of an individual machine as it pulls into synchronism immediately after connection to an in…nite bus will be treated later. Now the subscript s corresponding to the single stator winding of the elementary machine has been generalized to the subscripts fa. c indicate armature (stator) phase variables and the subscript f indicates …eld (rotor) variables. Self.or mutual inductance of windings will not vary with me (in this uniform-gap machine) if both subscripts are from the set fa. b. b..1. In a generator loss of synchronism (pull-out) can result in dangerous overspeed operation so sensors are usually applied to shut down the machine.. 5.3. I. b. Clearly the maximum jT j available occurs for sin RF = 1 hence RF = 90 : Synchronism will be lost if the machine torque needs to exceed this value to retain synchronism against the load.2) 4 c 5 = 4 Lca Lcb Lcc Lcf 5 4 ic 5 Lf a Lf b Lbc Lf f if f . inductances (or parameters) which are not a function of me will be denoted by an italic L rather than a script L: Note that equation 3. Laf . Mutual inductance between windings will vary with me if one subscript is from the set fa.e. Laa ..6 in the aforementioned notes. A sudden (e. cg which correspond to three stator phase windings.2 are given in ux matrix form by: 2 3 2 32 3 Laa Lab Lac Laf ia a 6 b 7 6 Lba Lbb Lbc Lbf 7 6 ib 7 6 7 6 76 7 (3.2 is analogous to equation 1. In this case the windings lie on opposite sides (stator side vs.g. a damped mechanical oscillation of the rotor about its new steady-state torque angle..

82 CHAPTER 3.5) 3. Note that a derivation of only one result (e. SYNCHRONOUS MACHINES of the elementary machine has been changed to f to denote a single rotor …eld winding. Lcc = Lcc = Lcc0 + Lcl .. the mutual inductance derives from the fraction (including sense or "polarity") of the total (self-inductive) ‡ of one phase winding which links to another phase winding. in view of the above discussion.3) (3.4) (3. ux .7 Self-Inductances The stator self-inductances Laa . and Lf f = Lf f = Lf 0 + Lf l : (3.6) Here the appended subscripts 0 and l denote the parts due to the spatial fundamental component and leakage..8) =+120 = (3. 3. 2 1 Laa0 : 2 (3. These mutual inductances are: Lba = Lab = Lab = [Laa0 cos ( )] Lca = Lac = Lac = [Laa0 cos ( )] and Lbc = Lcb = Lcb = [Laa0 cos ( )] =+120 =+120 = = 1 Laa0 . The corresponding decompositions are: Laa = Laa = Laa0 + Lal . Also note the simple relation between mutual inductance and self-inductance (e. and the other due to leakage. Lab ) su¢ ces because the remaining results follow by symmetry arguments. Lab and Laa0 ) which occurs because all windings are on the same (stator) core. Each of these selfinductances can be decomposed into two parts: one due to the spatial fundamental component of air-gap ‡ ux. Note that harmonic e¤ects due to stator or rotor slot openings are neglected here. I. Lbb .7) (3.e. Lbb = Lbb = Lbb0 + Lbl . respectively. (3. Lcc and the rotor self-inductance Lf f are not functions of me . The calculation of these inductances due to spatial fundamental components is given in Appendix B of the textbook. 2 1 Laa0 ..g.8 Mutual Inductances The mutual inductance between windings on the same side of the gap need be considered only for the stator because the rotor has only one winding.g.9) These results for mutual inductance are derived in Appendix B of the textbook.

We begin by writing out the phase-a ‡ expression: ux a = Laa ia + Lab ib + Lac ic + Laf if 1 Laa0 (ib + ic ) + Laf if 2 (3.15 for ia .11) (3. on a per-phase basis) by means of the so-called synchronous inductance which results under conditions of balanced (normally steady-state) three-phase armature currents.3. where we identify the constant synchronous inductance : 3 Ls = Laa0 + Lal : (3.15) = Ls ia + Laf if (3.14 reduces to a (3. In other words.16) by substitution of equation 3. + 120 ) : me These results for mutual inductance including calculation of Laf are derived in Appendix B of the textbook. It su¢ ces to derive this model for phase a because identical phase b and c models follow directly by symmetry arguments.13) which follows from the …rst row of the matrix equation 3. Laf ) su¢ ces because the remaining results follow by symmetry arguments. 3.17) 2 Here the factor 3 corresponds to the same factor in the forward stator mmf 2 wave which results under balanced conditions.8. The assumption (normally given in phasor form) that the three-phase armature currents are balanced is equivalent to an instantaneous balance in the time domain: ia + ib + ic = 0 ! ia = (ib + ic ) : Hence equation 3. . Again a derivation of only one result (e.2. 3. This yields a = (Laa0 + Lal ) ia + (3.3.e.9 Synchronous Inductance The operation of each phase can be analyzed separately (i.14) by substitution of the previously derived expressions for inductances from equations 3. and 3.g.10) (3. SYNCHRONOUS INDUCTANCE 83 The mutual inductance between windings on opposite sides of the gap are: Lf a = Laf = Laf cos ( and Lf b = Lbf = Laf cos ( Lf c = Lcf = Laf cos ( me me ) . (3..7..12) 120 ) . These mutual inductances are a function of me which is the necessary mechanism for electromechanical energy conversion to occur. mathematically the synchronous inductance incorporates the mutual (inductive) interaction between phase windings into a single per-phase model.9.

26) dt and …nally dia va = Ra ia + Ls + eaf (3. (3.27) dt by substitution of the previous equation into equation 3.27 and shown in Fig.21) (3. conversion of equation 3.27 to phasor form yields b b b b Va = Ra Ia + jXs Ia + Eaf (3. For AC steady-state operation the relation between voltage and current at the phase a armature terminals is now restated in RMS-phasor form. and the generated (speed) voltage eaf : Figs.3b show reference current directions for ia as inward and outward for conventional motor and generator operation.18) where the series DC armature resistance Ra is now included and the induced : voltage is ea = ddta . 5.3a and 5. I.25 we have dia ea = Ls + eaf (3. respectively.3. From equation 3. 5.23) Npoles =2 so dLaf = dt eaf = ! e Laf sin (! e t + ! e Laf If sin (! e t + e0 ) : (3.19 with substitution of equations 3.84 CHAPTER 3.20 and 3. SYNCHRONOUS MACHINES 3.10 where = ! Npoles 2 m = !e t + e0 . consists of three series elements: the resistance Ra .24) (3.18.28) .16 d d d : d a ea = = (Ls ia + Laf if ) = (Ls ia ) + (Laf if ) = eaa + eaf dt dt dt dt where dia : d (Ls ia ) = Ls eaa = dt dt dif dLaf : d eaf = (Laf if ) = Laf + if : dt dt dt Laf = Laf cos (! e t + me e0 ) (3. corresponding to equation 3.20) and (3.. the synchronous inductance Ls .22) By equation 3.10 Armature Equivalent Circuit The per-phase (line-to-neutral) equivalent circuit at the phase-a armature terminals will follow from: va = Ra ia + ea (3.e.19) (3. The per-phase equivalent circuit at the phase a armature terminals.25) This gives e0 ) by substitution into equation 3.21 with DC …eld excitation current if = If hence dif dt = 0: Now by equation 3.

34) is the e¤ective magnetizing reactance and : Xal = ! e Lal .25 for the generated voltage eaf . which corresponds to this decomposition.37) generated by the total air gap ‡ is called the air-gap voltage or voltage behind ux leakage reactance.4. we have Eaf = and hence the phasor representation b Eaf = jEaf ej e0 ! e Laf If p 2 (3.17 into equation 3. Refer to Examples 5. into Xs = X' + Xal . where : X' = ! e 3 2 Laa0 (3.36) is the leakage reactance. where max feaf g = ! e Laf If . Ia . is given in Fig.32) : (3. and the synchronous reactance is : Xs = ! e Ls : (3.) The phase a armature current is h p i ia = 2 Ia cos (! e t) (3. 5.4.2. Here X' corresponds to the forward fundamental spatial ‡ wave produced by the three armature phase currents under balanced ux conditions.3. Ia .35) (3. respectively.30) so b Ia = Ia ej0 (3.29. by substitution of equation 3. The typical order of magnitude of the impedance components (on a per-unit basis) is discussed on page 256 of the textbook. 5. . (3. The alternative form of the equivalent circuit.10.29) (Note that Xs is indicated for Ls in Fig.31) is the phase reference.1 and 5. Eaf denote phasors and their RMS magnitudes. 5. it is labelled in Fig. Eaf and Va . The voltage b : b b ER = jX' Ia + Eaf (3.33) The synchronous reactance decomposes. By equation 3. ARMATURE EQUIVALENT CIRCUIT 85 b b b where Va .3.

28 using the generator reference direction for current Ia . Refer to Example 5.38) Va = 2 by equation 3. under the condition Va = 0 (short-circuit armature terminals).5 for an illustration of the OCC including an air-gap line. Hence core losses can be determined by the di¤erence between machine loss measured with and without …eld excitation. Refer to Fig.28 so ! e Laf If p (3.40) b by equation 3. Va . The OCC starts linearly at the origin where saturation has not yet occurred and air-gap reluctance dominates the inductance Laf . The OCC is analogous to a magnetization curve for a DC machine because core saturation e¤ects can introduce a downward bend in the characteristic. Hence Laf = p 2Va ! e If (3. Under this condition b b b b Eaf = Ra Ia + jXs Ia = (Ra + jXs ) Ia (Ra + jXs ) = b Eaf : b Ia (3. which is normally used for comparative purposes.41) (3. SYNCHRONOUS MACHINES 3.86 CHAPTER 3. under the condition Ia = 0 (open circuit or no electrical load).11 Open-Circuit Characteristic The open-circuit characteristic (OCC) is Va as a function of If .12 Short-Circuit Characteristic The short-circuit characteristic (SCC) is Ia as a function of If . Under this condition Va = Eaf by equation 3. due to If acting alone. Refer to Fig. It relates the spatial fundamental of the air-gap ‡ to the mmf over the magnetic ux path. 5. In the latter case the base voltage is usually taken as the rated voltage of the machine. so Usually Ra Xs in which case the previous equation gives Xs ' Eaf . An OCC may use either absolute or per-unit values. here Va is varied by changing If : 3. Machine losses under open-circuit conditions (no electrical load) consist of the mechanical power necessary to overcome friction and windage at synchronous rotational speed plus core losses due to magnetic ‡ ux.39) can be determined by measurement.42) .6 for a typical curve of core loss vs. 5. Ia (3. The linear extrapolation or extension of the OCC from the origin is called the air-gap line.3.32.

47) p ! e Laf 2 ( jRa + Xs ) If (3. The linear SCC is derived as follows.g. b the air-gap voltage ER in equation 3.5.44) if Ra Xs : Refer to Fig.10 for a plot of short-circuit load loss. First we have jEaf ej by equations 3.32 for Eaf : Solution of this last equation for Ia gives ! e Laf If b Ia = p ej e0 . 0:15) of its normal value. 5.37 and hence the air-gap ‡ will be a small ux fraction (e.13 Measurement of Synchronous Reactance First.43) b by substitution of equation 3. 5.13. SCC is determined under short-circuit conditions created by application of a three-phase short-circuit with appropriate current sensors at the armature terminals of the machine. This implies that the machine is operating in an unsaturated condition. The …eld current is adjusted to obtain a range of . Refer to Fig.3. Refer to Example 5. Hence the I 2 R losses due to armature current can be found as the di¤erence between machine losses under short-circuit and open-circuit conditions. Core loss is considered negligible because the magnetic ‡ level is low ux under short-circuit conditions.33 and 3. MEASUREMENT OF SYNCHRONOUS REACTANCE 87 so Xs can be determined from Eaf on the OCC and Ia on the SCC. 263-65 of textbook for a complete discussion of these losses. Noting that equation 3. (3. Refer to pp. where the …eld current If must be the same for both.46) b = (Ra + jXs ) Ia (3.9 which illustrates OCC and SCC.40 so j ! e Laf If p 2 e0 b = (Ra + jXs ) Ia ej e0 (3. SCC is linear if the machine is unsaturated.45) 2 ( jRa + Xs ) hence the linear relation Ia = or 1 Ia = p 2 ! e Laf Xs If (3.40 expresses KVL around the loop where Va = 0 (instead of the machine rating). 3. Machine losses under short-circuit conditions consist of the mechanical power necessary to overcome friction and windage (determined from an open-circuit test with If = 0) at synchronous rotational speed plus losses due to armature current. In other words the armature reaction ‡ wave is moving with and opposing the ‡ wave due to the rotating …eld ux ux winding..

5. SYNCHRONOUS MACHINES armature current Ia not to exceed the machine rating.48.. Refer to pp. Refer to Fig.49) Substitution for if in equation 3. we will show SCR = 1 Xs. the armature ‡ wave has only a forward comux ponent which rotates in synchronism with the rotor given balanced three-phase . with if determined under the numerator and denominator conditions by equations 3. 3. 5. 261-62 and Fig.15 Field Winding Flux The …eld winding ‡ linkage is constant under DC …eld excitation assuming ux steady-state (synchronous) machine operating conditions with balanced threephase armature currents.e. OCC is determined under open-circuit conditions by measurement of Eaf vs. p er unit : a Here Va and Ia were taken as rated values for the machine so Va is the I impedance base for conversion to per unit impedance. 3.47 respectively. 261-62 of textbook for interpretation of Xs under saturated conditions in machine.9 which illustrates OCC and SCC for the determination of Xs : Refer to pp. If . This process is analogous to determination of the series impedance in a Thevenin equivalent circuit. gives 2)Va ! e Laf p ( 2)Ia ! e Laf ( p SCR = = Xs 1 Xs a ( Va ) I = 1 Xs. I. SCR = [if ]SCC:Ia at rated value | {z } "AFSC" (3. respectively. The unsaturated synchronous reactance Xs follows from equation 3.48) where "AFNL" and "AFSC" mean "amperes …eld no load" and "amperes …eld short circuit". Refer to Example 5.14 Short-Circuit Ratio "AFNL" The short-circuit ratio (SCR) is }| { z : [if ]OCC:Va at rated value .4.42 by substitution of the values for Ia and Eaf read from the SCC and the OCC at the same value of If : Eaf must be read from the air-gap line on the OCC because the SCC is under unsaturated conditions.38 and 3.9 for a discussion of saturated synchronous reactance. which is not surprising as the armature equivalent circuit is already in Thevenin form.88 CHAPTER 3. p er unit : (3.

The external system may contain more synchronous machines. transmission lines.50 into equation 3. STEADY-STATE POWER-ANGLE CHARACTERISTIC 89 armature currents. 5.51) which serves as the phase reference. an initial (RL) transient occurs upon application of the DC …eld excitation.12. In phasor terms. the "sending" voltage is (3. as shown in Fig. ux A voltage can be induced into the …eld winding under transient unbalanced conditions on the armature currents. Here a synchronous machine with known armature (Thevenin) equivalent circuit is connected to an external system represented by another Thevenin equivalent circuit.16. and transformer banks.11a. This result is equivalent to that which would be obtained by writing out and evaluating the …eld ‡ expression for f from equation 3. 5. Machine performance can be a¤ected under these conditions.52. b This problem is a special case of power ‡ between two voltage sources E1 ow b and E2 through a series impedance Z = R + jX = jZj ej Z ! Z : = arctan X R . Z b E2 (3.55) . Here Z represents the series combination of Thevenin equivalent impedances of the machine and the external system.51.53: j E2 E1 ej( b E1 e I= = j Z jZj e jZj Z) E2 e j jZj Z (3. the "receiving" voltage is b : E2 = E2 ej0 = E2 .2.53) b so we will …nd I to determine I cos ( ) : By substitution of equations 3. and 3.16 Steady-State Power-Angle Characteristic We determine the maximum power that can be delivered by a synchronous machine without loss of synchronism due to excess torque requirements. b : E1 = E1 ej .52) and the series-element current is with phase angle receiving end is determined by the network. 3.54) b E1 b: I = I2 ej = (3. The real power delivered to the P2 = E2 I cos ( ) (3.3. Hence the …eld ‡ linkage due to the armature currents is ux constant (non time-varying) and the armature currents do not induce a voltage in the …eld winding.50) as shown in the network of Fig. (3. such that the …eld current settles to its …nal DC value V If = Rf : f 3. Also.

64) occurs at = 90 if E1 and E2 are constant.63 yields P = Eaf VEQ sin ( ) XS + XEQ (3.62) if R jZj because that implies jZj ' X and Z ' 0: Equation 3.56) This yields = Z) (3.59) by equation 3.54 with substitution of equation 3.63 is called the power angle characteristic and is called the power angle for synchronous machines. or alternatively P2 = where Z 2 E2 R 2 jZj R X (3.57.58) E1 E2 cos ( jZj E1 E2 sin ( + jZj Z) jZj Z) . Maximum power transfer P1 ' P2 ' E 1 E2 X (3. SYNCHRONOUS MACHINES = E1 cos ( jZj E1 cos ( jZj Z) Z) E2 cos ( jZj E2 R jZj 2 Z) : (3.60) : = 90 Z = arctan : (3.63) E1 E2 sin ( jZj Z) + 2 E1 R jZj 2 : (3.57) = cos ( Z) = R : jZj 2 E2 R 2 (3.61) Likewise the power delivered by the sending end is P1 = It is important to note that P1 ' P2 ' E1 E2 sin ( ) X (3.65) . ow For the synchronous machine shown in Fig. (3. equation 3. 12. Further E1 leads (lags) E2 if > 0 ( < 0) and power ‡ is from E1 to E2 (E2 to E1 ).90 which gives n o b Re I = I cos I cos because cos ( Finally P2 = CHAPTER 3.

power. and XEQ ! X: However Va ! E1 is not constant as P changes. Refer to p. Compounding curves relate If to Ia under the constraint that Va is constant as load power P varies.17. Refer to pp. However Eaf is ultimately limited due to thermal factors regarding the …eld current If : Also generators are usually operated at a power angle signi…cantly less than 90 to provide adequate stability margins. any restriction on Ia is removed now.67) (3.6 and 5. Next assume the limiting value If which implies a certain limiting value Eaf which we will use instead of If . Hence jSj = P 2 + Q2 = constant 2 (3.66) which implies jSj = jVa j jIa j = constant. due to simultaneous limits on Ia and If imposed by heating of armature and …eld windings respectively. 276-78 of the textbook for a discussion and Figs. VEQ ! E2 . and (XS + XEQ ) ! X: Recall that this is a per-phase analysis so total power is PT = 3P: It is possible to apply equation 3.7. 5. Va ! E2 . respectively.17 Steady-State Operating Characteristics Steady state operating characteristics refer to relations between Va . 275 of textbook for additional details and Fig. STEADY-STATE OPERATING CHARACTERISTICS 91 by the identi…cations Eaf ! E1 .68) so this locus is a circle in the P Q plane with center at the origin and radius jSj KVA. VEQ ! E2 . Capability curves give the locus of maximum real power P and reactive power Q for a machine operating at rated voltage Va . We have b b b Va = jXs Ia + Eaf (3.3.63 to di¤erent subsections of the network by using the alternative identi…cation Eaf ! E1 .65 indicates that power transfer can be increased by increasing Eaf if VEQ is a …xed system-bus voltage. 3. A brief derivation or construction of the capability curve follows with reference to Fig.16 and 5.17.69) .65 and the torque in equation 3.17 which illustrate a typical capability curve and its construction. Ia . Refer to Examples 5.1 have the same form with respect to the power angle : This con…rms that the limiting factor in power transfer is the maximum available torque. and e¢ ciency. XS ! X or the alternative identi…cation Va ! E1 . Equation 3. 5. with various …xed power factors as a parameter. power factor. First assume the limiting value Ia (a magnitude without b constraint on the phase of Ia ) and operation at rated Va : Recall S = P + jQ = Va I a (3. so an additional relation would be required to relate Va to P: The power-angle characteristic in equation 3.15 which illustrates typical compounding curves. If . 5.

2 Va Xs Va Eaf Xs : Q plane and radius This locus is a circle with center at Va Eaf Xs in the P : The valid operating region is the area in the P Q plane where both constraints are satis…ed simultaneously. hence b b j Va Eaf b b Va Eaf b Ia = = : (3. Note that the machine rating (apparent power and power factor) is commonly speci…ed as the intersection of the two circles.71) which gives = by simple algebraic rearrangement. This region is the intersection of the areas enclosed by the two circular loci with the added requirement P > 0.72) (3.28 using the generator reference direction for Ia and assuming Ra = 0. as shown in Fig.70) jXs Xs b We assume Va = Va is the phase reference here and compute: 2 3 b b j Va Eaf 2 b jVa Eaf b b b 5 = jVa P jQ = Va Ia = Va 4 Xs Xs Xs P j Q+ 2 Va Xs (3.17.72 yields P2 + Q + = 0.73) where the magnitude Eaf is …xed or limited but the angle is arbitrary. 5. Refer to pp. V curves relate If to Ia under the constraint that Va and load power P are constant with various …xed power factors as a parameter.74) so taking the square of the magnitude of both sides of equation 3. SYNCHRONOUS MACHINES b from equation 3. . 5. However b jVa j Eaf b jVa Eaf Va Eaf = = Xs jXs j Xs 2 Va Xs 2 2 (3. Note that b Eaf = (Eaf ) ej b jVa Eaf Xs (3. 278-79 of textbook for additional details and Fig.92 CHAPTER 3.18 which illustrates typical V curves.

In either case there is only a forward-rotating armature ‡ ux-density wave if stator currents are balanced.Chapter 4 Polyphase Induction Machines 4. In an induction machine the rotor currents (hence "operating …eld") are generated by induction (generalized transformer action) from the armature ‡ ux-density wave due to slip (di¤erence in rotational speed of the ‡ ux-density wave and rotor winding).3 of the notes entitled Rotating Machines. A wound rotor is wound with the same number of poles as the stator whereas a squirrel-cage rotor will automatically acquire that number of poles by induction from the stator ‡ ux-density wave. Many principles for analysis of an induction machine carry over from analysis of a synchronous machine. A wound 93 . However the rotor ("…eld") winding(s) of an induction machine and a synchronous machine are fundamentally di¤erent by design: the former are loops closed by either a short circuit or external resistance (hence no source applied). 4.1 Introduction A detailed analysis of AC polyphase (three-phase) non-salient pole induction machines will be presented here.2 Rotor Construction Wound rotors and squirrel-cage rotors are the two possibilities in induction machines. The number of rotor phases is often but not always the same as the number of stator phases. whereas the latter are driven by an external source to establish the operating …eld. This is the fundamental mechanism by which electromechanical energy conversion occurs. Some underlying principles of induction machines were given previously in Section 1. The stator (armature) windings of three-phase induction machines and three-phase synchronous machines are very similar if not identical.11.

to produce a more favorable torque-speed characteristic.94 CHAPTER 4. Hence squirrel-cage rotors are by far the most common.1) The fractional slip is where ns is the synchronous rotational rate (RPM) of the armature ‡ ux-density wave and n is the mechanical rotational rate (RPM) of the rotor. the last equation yields ! r = ! s ! m = s! s (4. internally short-circuited.4) : !r = sfe (4. Its major disadvantage is cost and reduced reliability due to the presence of slip rings. 6. Denoting the slip frequency of the rotor by ! r = s! s .1. POLYPHASE INDUCTION MACHINES rotor allows connection of external series resistance through slip rings whereas a squirrel-cage rotor does not because it is permanently.20 of the 2 2 aforementioned notes di¤ers from that used here.2. In some squirrel-cage rotor designs electromagnetic …eld phenomena are used to vary the internal rotor resistance automatically with speed. hence rotor ‡ ux-density wave. (4. Note that the rotor ‡ ux-density wave rotates at fs when viewed from the . respectively. 4.3 which show machines designed with wound and squirrel-cage rotors. Refer to Figs. The rotor currents created by these induced voltages will cause a rotor mmf wave.2) from which it follows that ! m = (1 s) ! s (4. and 6.) or 4.5) fr = 2 : : where fe = fs = !s and fm = !m : (The notation in equation 1. The advantage of a wound rotor is that external resistance can be used to control certain motor characteristics such as torque versus speed.3) where ! s and ! m are the synchronous (magnetic ‡ ux-density wave) and mechanical (rotor) angular frequencies (where one revolution corresponds to 2 : radians). 6. which advances (in the same direction as the stator ‡ ux-density wave) around the rotor at frequency fr .3 Slip : ns n s= ns (4.4 Rotor Induction Rotor voltages induced by the rotating stator ‡ ux-density wave (Faraday’ law) s will have electrical frequency fr (which corresponds to ! r ). because that is the frequency at which the stator ‡ appears to vary when viewed from the rotor ux (due to relative motion). This equation gives n = (1 s) ns .

2 | {z } | {z } N2 21 (4.4 irrespective of any particular value of slip s: Hence the rotor and stator …elds (‡ ux-density waves) rotate in synchronism.8) = L21 i1 + L22 i0 .7) N2 22 The mutual ‡ ux 21 . GENERALIZED TRANSFORMER 95 stator because fr + fm = fs by equation 4.e. 1l = 11 21 and 2l = 22 21 are the the stator and rotor leakage ‡ uxes. I. and the turns ratio is : N1 n= : N2 (4. Consider the stator and rotor ‡ linkages (for one phase): ux 1 and 0 2 = L11 i1 + L12 i0 2 | {z } | {z } N1 11 (4.5 Generalized Transformer Transformer analysis is generalized here to admit relative motion between the secondary (rotor) and primary (stator) windings.10) give the stator and rotor voltages. respectively. where 12 = 21 . respectively.12) .6) N1 12 where unprimed and primed variables indicate stator and rotor quantities.. and e1 = e0 = 2 d 1 = dt d 2 = dt N1 d 11 dt 21 + N1 d 12 dt 22 (4. (4. corresponds to the rotating sinusoidal : : stator ‡ ux-density wave in the gap. 1. the basic transformer equivalent circuit is applied with some external modi…cations in the secondary loop to account for slip (relative motion). where all phases operating simultaneously under balanced conditions will produce a net forward ‡ ux-density wave (and no backward ‡ ux-density wave).9) N2 d dt + N2 d dt .4.11) n and b0 b I2 = nI2 : (4. This situation is modeled by the standard transformer equivalent circuit in Fig. where the extracted n : 1 ideal transformer represents the e¤ects of the mutual ‡ ux 21 which imposes the constraints: 1 b b 00 E2 = E2 (4. 4. We consider the forward stator ‡ uxdensity wave due to each phase separately (per-phase analysis).5.

13) b0 b 00 E2 = sE2 (4. Substitution of this last expression into the previous equation yields Z2 = n2 s 0 (R2 + js! e L0 ) = 2l 0 n2 R2 R2 + jn2 ! e L0 = + jX2 2l s s (4. 1.16.20) (4. Note that X2 is proportional to slip s due to the b0 change in the frequency of the induced voltage E2 : The equivalent impedance seen on the stator (primary) side of the ideal transformer is b : E2 (4. The e¤ect in equation 4. s b I0 b I2 = 2 n (4.17) Z2 = b I2 where by equations 4. The impedance 0 : 0 0 Z2 = R2 + jX2 (4.12. where 0 : 0 Z2 = R2 + js! e L0 2l b0 E2 = b I0 2 n2 s 0 Z2 (4.18) (4.96 CHAPTER 4.14 is represented by the multiplier s inserted at the secondary of the ideal transformer in Fig. and n b0 b E2 = E2 .16) 0 due to the rotor leakage inductance L0 .22) .21) by equations 4.10: d so by Faraday’ law.14) (4. where R2 and L0 are physical constants 2l 2l 0 due to the rotor construction.14.11 and 4.15 and 4.19) by equation 4.15) 0 on the rotor side represents the extracted rotor resistance R2 and rotor reactance 0 X2 = ! r L0 = s! e L0 2l 2l 21 dt !s d 21 dt (4. So Z2 = n s b0 I2 n b0 E2 = n2 s by substitution of the previous two expressions into equation 4. POLYPHASE INDUCTION MACHINES The e¤ect of slip between rotor and stator is to modify the time derivative of 21 which appears in equation 4.17. where fr = sfe or ! r = s! e is the frequency of the induced s b0 rotor voltage E2 .

6 and 4. Three stator phase windings spatially displaced by 120 are driven by balanced three-phase currents. 6.7. respectively. EQUIVALENT CIRCUIT where and 97 : 0 R2 = n2 R 2 : X2 = n2 ! e L0 : 2l (4. On the other hand a squirrel-cage rotor is a di¤erent con…guration which may be viewed as a collection of single conductor loops distributed at uniform angular increments around the rotor. The equivalent stator (primary) current follows by integration over the current sheet so the general form of Z2 will be the same as before. These loops happen to be joined by the end plates which renders their operation similar to a parallel rather than a series connection.7 and 6. Note that resistances R1 and Rc have been included to represent primary winding resistance and core losses. This is shown in Fig. 1. equations 4.23) (4. The net e¤ect of all these individual loop currents induced by the rotating ‡ ux-density wave is to approximate a sinusoidal current sheet. where turns are inherently in series.6 Equivalent Circuit The per-phase equivalent circuit in a …nal form follows by inserting Z2 from equation 4.7 Current and Flux First we review the process by which the stator ‡ ux-density wave is produced.7 can be interpreted by using equivalent turns N1 and N2 which represent the net e¤ect of the current sheet induced across the winding distribution by the rotating ‡ ux-density wave . The preceding analysis can be applied to either wound rotors or squirrelcage rotors with appropriate interpretation.22 for the equivalent impedance seen from the stator side in Fig. as slip s varies. so the corresponding mmf wave lags the rotating current sheet by 90 : This mmf wave drives .4. Equation 4.9 which is a development from Figs. This equivalent circuit can be applied to determine many characteristic relations in induction machines between current. speed. 4. 6. For a wound rotor. The application of Ampere’ law to determine s mmf amounts to integration of the sinusoidal current sheet. torque.22 for Z2 indicates that the reactance remains constant as X2 while only the resistance R2 changes.8. etc.6 and 4. Each phase winding is individually distributed to approximate a sinusoidal current sheet and the net e¤ect is to produce a single forward rotating sinusoidal current sheet. One loop can be analyzed directly starting from equations 4. 4.6. losses. s 0 b0 The former is due to o¤setting e¤ects: E2 decreases with slip while X2 also decreases with slip.24) Here R2 and X2 are the resistance and reactance seen in Z2 (stator side) when the rotor is stationary (s = 1).

so it will lag the current sheet by 90 .28) rs = is the angle between the rotor rotating ‡ ux-density (B …eld) wave and the stator rotating ‡ ux-density (B …eld) wave. 6. (4.10 and 4. It is su¢ cient to consider only one of the induced phase voltages e0 to determine the phase by which the rotating voltage 2 wave leads the rotating current sheet. Now consider the phase of the rotating current sheet produced in the rotor by the induced rotating voltage wave.5 and 6.98 CHAPTER 4. (The latter statement follows because equations 4.17. The angle rs follows by accumulation of the phase shifts which are incidental to the process described above.10. ! X2 X2 = arctan s : (4. Hence when slip s is small 90 : (4. The rotor rotating voltage wave due to the balanced set of three induced rotor phase voltages e0 is in phase with the ‡ ux-density wave but with opposite polarity.9.27) = arctan R2 R2 s Finally the rotor rotating mmf wave is given by integration of the sinusoidal current sheet.22 or from the equivalent circuit in Fig. In cases where the resultant air-gap rotating ‡ density wave ux is approximately constant (4. POLYPHASE INDUCTION MACHINES the sinusoidal mutual stator ‡ ux-density wave in the gap without any further 11 21 phase shift.9: (4.29) rs Angle rs becomes more negative than 90 if leakage reactance X2 and/or slip s increase. respectively.6 which are pertinent to torque angle considerations for wound rotors and squirrel-cage rotors. to produce d dt and most of d dt in equations 4. where the torque angle r is the angle between the rotor rotating ‡ ux-density wave and that resultant wave.30) r rs . equivalently we can write Z2 in polar form: Z2 = jZ2 j ej : The phase 6. Refer to Figs. by 2 equation 4. upon which an incidental sign or polarity change is overlaid.18 and 4.19 indicate that these primed and unprimed voltage-current pairs are b b related by simple scaling. This means that the rotating rotor ‡ ux-density wave. which is driven directly by the rotor mmf wave. . The phase will be the phase between the b0 b0 individual rotor voltage E2 and the individual rotor current I2 or equivalently b b the phase between E2 and I2 . We conclude that (90 + ) (4. will lag the current sheet by 90 .25) by equation 4.) The phase between E2 and I2 is simply the angle of the impedance Z2 : = \Z2 .26) follows from equation 4.

32) by equation 4.34) .g.. which means fr is in the range of 1 Hz to 6 Hz for a 60 Hz motor.9 4. In steady-state operation the torque angle r is constant so the torque will be steady or constant in an induction machine.31) which is equation 1.28 and 4.31 is applied to an induction machine by assuming: the magnitude sr of the resultant air gap ‡ ux sr is approximately constant for a given applied stator voltage and frequency. speed curve .33) KIr sin r. Equation 4.4. In such region Z2 is predominately resistive b0 so Ir is increasing with slip because induced rotor voltage magnitude E2 is increasing with slip.9.8. in Section 1. This led to the expression for torque: T = Npoles 2 2 sr Fr 2 sin r. Hence we can rewrite this equation as T = where K is a constant. in view of equations 4.29 if slip is small. based on equation 4. Likewise sin r decreases as the phase angle of Z2 increases due to its reactive component. The derivation was based on the assumption that the stator and rotor ‡ ux-density waves rotate in synchronism with each other. (4. under normal operating conditions slip is 2% to 10% in a typical squirrel-cage motor.7 of previous notes entitled Rotating Machines. This produces a steep nearly linear rise starting at zero slip in the torque vs. which is true for induction machines as demonstrated above. (4. and rotor mmf Fr is proportional to rotor current Ir .18. TORQUE 99 4. E. 6.8 Torque Torque was derived from coenergy based on the magnetic …eld in the gap.81).99 in the aforementioned notes (same as textbook equation 4. as illustrated in Fig.32. 4.1 Analysis by Equivalent Circuit Power and E¢ ciency Total (real) power transferred across the gap is 2 Pgap = nph I2 R2 s (4. Torque will reach a maximum (called breakdown torque) and then decrease as slip increases further for two reasons. The increase in Ir abates as Z2 asymptotically approaches a constant reactance.4. which gives T = KIr (4.30.

100

CHAPTER 4. POLYPHASE INDUCTION MACHINES

with reference to the equivalent circuit in Fig. 6.9, because this is the power transferred through the terminals a-b where the rotor equivalent circuit appears with reference to the stator. The power dissipation in the rotor resistance is
2 Protor = nph I2 R2

(4.35)

which can be deduced either by referring I2 and R2 to the rotor side or by observation that equation 4.34 reduces to this expression is the rotor is stationary (s = 1). The mechanical power developed by the motor is Pmech = Pgap
2 = nph I2 2 Protor = nph I2

R2 s

2 nph I2 R2

(4.36)

R2 s

2 R2 = nph I2 R2

1 s

s

:

This gives Pmech = (1 and Protor = sPgap (4.38) by comparison with equations 4.34 and 4.35, respectively. Hence the fractional power e¢ ciency is : Pmech =1 s (4.39) = Pgap so it is very ine¢ cient to operate an induction motor at large slip s: Noting that R2 s or R2 = R2 + R 2 s 1 s s (4.41) R2 = R 2 1 s s (4.40) s) Pgap (4.37)

from equation 4.36, it is useful to express R2 by equation 4.41 for power analysis s as shown in the alternative equivalent circuit of Fig. 6.10. Refer to Example 6.1 and Practice Problem 6.1.

4.9.2

Torque
Pmech = ! m Tmech (4.42)

Mechanical power is where Tmech is mechanical torque, so Tmech = Pmech Pmech = !m (1 s) ! s (4.43)

4.10. APPLICATION OF THEVENIN EQUIVALENT CIRCUITS where ! m = (1 s) ! s : This gives Tmech =
2 nph I2 Pgap = !s !s R2 s

101

(4.44)

by application of equation 4.37, then equation 4.34; recall that !s = 4 fe 2 = !e : Npoles Npoles

Actual power and torque available at the mechanical shaft must take friction, windage, and other mechanical losses into account. I.e., Pshaf t = Pmech Prot ;

where Prot represents such mechanical power losses, and Tshaf t = Pshaf t = Tmech !m Trot ;

where Trot represents such mechanical torque losses.

4.9.3

Examples

Refer to Example 6.2 and Practice Problem 6.2

4.10
4.10.1

Application of Thevenin Equivalent Circuits
Reduction of Motor Equivalent Circuit

b I2 must be evaluated in the equivalent circuit of Fig. 6.9 or Fig 6.10 to determine characteristics such as torque and power vs. parameters such as s; R2 ; and X2 : This evaluation can be simpli…ed, with additional insight as well, by substitution of a Thevenin equivalent for the network to the left of terminals a-b in Fig. 6.9. Recall the Thevenin equivalent from circuit theory. It requires determination of the open circuit voltage (with "loading" removed) and equivalent impedance (of "dead" network) seen looking back into the terminals a-b. By voltage division, the open circuit voltage is b b V1;eq = V1 jXm R1 + j (X1 + Xm ) (4.45)

where in most induction motors negligible error will result by neglecting the stator winding resistance R1 : The equivalent impedance is b Z1;eq = (R1 + jX1 ) k (Rc kjXm ) jXm (R1 + jX1 ) b Z1;eq = R1 + j (X1 + Xm ) (4.46)

or

(4.47)

102

CHAPTER 4. POLYPHASE INDUCTION MACHINES

where the input voltage source V1 is replaced by a short circuit and Rc has been neglected in the last equation. b Now I2 is evaluated by application of the Thevenin equivalent: b I2 = b Z1;eq + jX2 + b V1;eq
R2 s

:

(4.48)

4.10.2
Torque is

Torque

Tmech =

1 !s

"

2 nph V1;eq

R2 s 2

R1;eq +

R2 s

2

+ (X1;eq + X2 )

#

(4.49)

b from equation 4.44 with substitution for I2 from equation 4.48. Refer to Fig. 6.14 which shows a general plot of this expression for constant input voltage and frequency. Refer to Fig. 6.15 which shows torque, power, and current for the motor in Examples 6.2 and 6.3. Maximum torque (breakdown torque) follows from equation 4.44 by …nding the value of the resistance R2 for maximum power transfer from the remainder s of the network. It is well known in network theory that this maximum occurs if and only if the resistance equals the magnitude of the impedance seen looking into the remainder of the network. Hence we take q R2 2 2 = R1;eq + (X1;eq + X2 ) (4.50) s and solve for R2 smax T = q : 2 2 R1;eq + (X1;eq + X2 ) (4.51)

Substitution of smax T into equation 4.49 yields the maximum torque: 2 3 2 0:5nph V1;eq 1 4 5: q Tmech = 2 !s 2 R + R + (X +X )
1;eq 1;eq 1;eq 2

(4.52)

4.10.3

Examples

Refer to Examples 6.3 and 6.4.

4.11
4.11.1

Parameter Determination
Introduction

Motor parameters can be determined by a combination of no-load, blocked rotor, and DC resistance measurements. No-load I 2 R losses in motors are higher than

The rotational losses due to friction and windage are 2 Prot = Pnl nph I1 R1 (4.58) where Sbl = nph V1 I1 Then the blocked-rotor reactance corrected to rated frequency fr is Xbl = fr fbl Qbl 2 : nph I1 (4.11. current I1 .54) d! m (4. the motor voltage V1 (line-to-neutral). to simulate s = 0:25 the test frequency should be 0:25fe and result at fe can be found by backward extrapolation. 4.55) dt from physical laws. …nd Prot from power-o¤ decay transient: J or d! m = dt 0 Prot = Trot = Prot !m (4.3 Blocked Rotor Test Operating with blocked rotor. the equivalent circuit is identical. Alternatively. Hence abruptly switch o¤ power and measure d!m to obtain dt 0 Prot which now does not include core losses. Stator resistance should be measured separately with a DC ohmmeter. current I1 . A blocked rotor test is analogous to a short-circuit test of a transformer.g.2 No-Load Test Operating at no load.4.56) !m J Also observe that Xnl X11 = X1 + Xm R2 s (4.60) (4. E.53) where rotor I 2 R losses can be neglected because the rotor current is very small at no load. First determine blocked-rotor reactive power: q 2 2 Qbl = Sbl Pbl (4. and input power Pnl are measured. the motor voltage V1 (line-to-neutral).. This allows determination of core losses: 0 2 Pcore = Pnl Prot nph I1 R1 : (4.11. and input power Pbl are measured.57) is very large for because impedance of paralleled branch containing X2 + small s: 4.11. Voltage and frequency can be adjusted to simulate actual operating conditions in the rotor due to slip. PARAMETER DETERMINATION 103 I 2 R losses in transformers under open circuit test conditions because motors require greater excitation current due to the reluctance of the air gap.59) .

5 and 6. .6. 6.68) Xbl 2 : (4.63) ! X1 + 2 Xm R2 + Xm (Xm + X2 ) 2 (4.64) 2 R2 + (Xm + X2 ) from the equivalent circuit.11 with s = 1: The input impedance is Zbl = R1 + jX1 + (R2 + jX2 ) k (jXm ) ! 2 Xm +j = R1 + R 2 2 2 R2 + (Xm + X2 ) (4.67) Now these equations can be solved for X2 = (Xbl and R2 = (Rbl R1 ) Xm + X2 Xm X1 ) Xm Xm + X1 (4.69) where X1 + Xm is known from the no-load test.4 Examples Refer to Examples 6.62) to the input impedance of the equivalent circuit in Fig. 4.66) Xm + X2 and Xbl = X1 + X2 Xm Xm + X2 : (4. This reduces to Zbl = R1 + R2 Xm Xm + X2 2 + j X1 + X2 Xm Xm + X2 (4. POLYPHASE INDUCTION MACHINES The blocked-rotor resistance is Rbl = Pbl 2: nph I1 (4.61) The equivalent circuit parameters can be determined by matching the measurment Zbl = Rbl + jXbl (4.104 CHAPTER 4.11.65) by taking approximations such as R2 Xm : Matching resistive and reactive components from this last equation to Zbl gives: 2 Xm Rbl = R1 + R2 (4.

.

where s is the rotor slip with respect to the forward-rotating wave.rotating) component(s). Recall that stator winding currents are balanced (unbalanced) if the net stator mmf wave has only a forward-rotating (both forward.and two-phase induction motors is closely related to the analysis given in the previous notes entitled Polyphase Induction Machines.1) .and backward.2) ae ] [cos ! e t] (5.Chapter 5 Single.and Two-Phase Induction Motors 5. Analysis under unbalanced conditions is achieved by superposition of the separate e¤ects of forward.rotating mmf waves. I. With respect to the backward-rotating wave.1 Single-Phase Induction Motor Decomposition of mmf Wave The fundamental sinusoidal mmf for a single phase winding is Fa ( ) = Fmax [cos where ia = Ia cos ! e t. On the other hand stator currents are inherently unbalanced in a single-phase motor and may be unbalanced or balanced in a two-phase motor.2 5. The analysis for polyphase (three-phase) induction machines was based on the assumption that stator currents are balanced. the e¤ect of each wave is determined by the same method used to …nd the e¤ect of the forward-rotating wave in the analysis of the polyphase induction machine. 105 (5. 5.2. it is only necessary to apply the fact that the rotor slip becomes 2 s.and backward..1 Introduction Analysis of single.e.

28c.10) Here the subscript a refers to phase a and the superscripts + and refer to forward.2 Equivalent Circuit The stator equivalent circuit for steady-state operation with a forward mmf wave was derived in Sections 1. ae (5.6) (5.4) Note that here Fa ( ) remains …xed in space with the envelope cos ae while its amplitude is pulsating with respect to time due to the factor cos ! e t: Refer to Fig. However now the peak mmf Fmax in equation 5. This equivalent circuit gives the b b constraint between the stator voltage V and the stator current I which generates b the forward wave.6 of the previous notes entitled Polyphase Induction Machines. 4.and backward.17. it is shown in Fig.9) and backward-travelling wave is : Fa = Fmax 2 [cos ( ae + ! e t)] : (5. This was based on the assumption that no backward wave was present because of balanced operation of the polyphase machine.8) where the forward-travelling wave is + : Fa = Fmax 2 [cos ( ae ! e t)] (5.3) and : Fmax = kw Nphase Npoles Ia : (5.7) (5.1 where we take = and = ! e t: This yields + Fa = Fa + Fa ae (5. SINGLE. where V is the sum of voltage induced by the forward wave and the voltage drop across the stator leakage impedance. respectively.5 and 1. it is illustrated in phasor form by Fig.2 of the previous notes entitled Rotating Machines. The decomposition of Fa ( ) into the sum of forward and backward travelling waves is accomplished simply by applying the trig identity cos cos = 1 2 [cos ( ) + cos ( + )] (5. 5. 6.9 2 .106 CHAPTER 5. This decomposition of the mmf wave due to a single phase was given in section 1.travelling waves.AND TWO-PHASE INDUCTION MOTORS : Npoles = 2 4 a. 4.5) to equation 5.2.28.9.

I. respectively. so V must be scaled 1 by the factor 2 .12) Likewise the impedance of the equivalent circuit for the backward wave must 1 be scaled by the factor 2 : The net equivalent circuit.11c) indicates that the magnitude of the forward ‡ ux-density wave will increase relative to that of the backward ‡ ux-density wave as slip increases. so the relative rotational rate between the rotor and the backward wave is !m ( ! s ) = (1 s) ! s ( ! s ) = (2 s)! s : | {z } slip (5. The resultant torque is the di¤erence of forward and backward torques. 2 The stator equivalent circuit for steady-state operation with a backward mmf wave is essentially the same as that for a forward wave.b in the equivalent circuit of Fig. illustrated in Fig.e.f > Emain. the latter behavior would be equivalent ux to superimposing separate forward and reverse torque-speed characteristics of a polyphase induction motor driven by a constant voltage in each case.11c).b between the induced voltages due to these waves. based on the decomposition in equation 5. I. The resultant torque-speed characteristic. SINGLE-PHASE INDUCTION MOTOR 107 b is one-half of the value used in the previous derivation.. This occurs because the rotor currents induced by the forward and backward ‡ ux-density waves have the e¤ect of increasing and decreasing the forward and backward mmf waves. Hence the torquespeed characteristic of the single-phase machine is more favorable than would be expected from superposition of characteristics corresponding to constant forward and backward ‡ waves. The magnitudes of these forward and backward waves can be determined b b from Emain. as seen from the zero value of the resultant torque-speed characteristic at ! m = 0 (s = 1)..2b). follows from the scaled individual equivalent circuits for the forward and backward waves. b b because this leads to the relation Emain. whereas the single stator leakage impedance is common to both waves. This is expected because the single pulsating stationary .2. with the exception that the rotor slip s is replaced by 2 s: This new value of slip follows because ! m = (1 s) ! s (5.e.f and Emain. 9. There is no starting torque in an induction motor with only one (main) stator winding.b ) follow from the previous analysis of polyphase induction machines. 9. 9. the portions of these individual equivalent circuit impedances which represent voltages induced in the stator (primary) winding by forward and backward ‡ ux-density waves are placed in series because these induced voltages are additive in the winding.11c). This is accomplished simply by scaling the impedance of the stator equivalent circuit by the factor 1 .11) and the backward wave is rotating at speed ! s . The equivalent circuit in Fig.5.f and Emain.8 and shown in Fig. The forward and backward torques corresponding to these waves (hence corresponding to b b these induced voltages Emain. 9. accounts for the magnitude variation of forward and backward ‡ ux-density waves with slip.

b accounts for one-half of the power dissipation in R2 with .and backward. (5. The introduction of a second stator winding to obtain starting torque is discussed in Section 5.f = I 2 (0:5Rf ) (5.11c) for the forward.f if the rotor (secondary) were open circuited. A closer examination will show that Pgap. 2 s The forward and backward impedances are (5.AND TWO-PHASE INDUCTION MOTORS spatial mmf distribution was resolved only mathematically into forward and backward mmf waves. However a net torque is developed once the rotor is moving forward because the e¤ect of the forward ‡ ux-density wave becomes larger than that of the backward ‡ ux-density wave.b = I 2 (0:5Rb ).108 CHAPTER 5.1 Analysis by Equivalent Circuit Forward and Backward Impedance : Zf = Rf + jXf = R2 + jX2 k (jXm ) s R2 + jX2 k (jXm ) . SINGLE. 9. s which has been referred to the stator (primary) side.3. This follows from examination of Fig. All of current b Imain would enter the magnetizing reactance 0:5Xm. the forward equivalent circuit su¢ ces for this discussion. due to the di¤erent rotor slips relative to the forward and backward waves. 9.13) and : Zb = Rb + jXb = (5. These are evident from Fig. 5.4.16) respectively.11c).3 5. Physical insight is gained by reviewing the interpretation of the (generalized) transformer equivalent circuit which underlies the operational equivalent circuits in Fig.2 Power The powers delivered by the stator winding to the forward and backward …elds are Pgap. 9. shunts the magnetizing inb ductance and diverts a signi…cant fraction of Imain away from this inductance. 5.15) and Pgap. Conux b sider the input current Imain which enters either of these operational equivalent circuits.14) respectively. This is an equivalent behavior with respect to the ‡ ux-density wave (magnetic b …eld) whereas physically all of Imain ‡ ows through the magnetizing inductance (primary).3.‡ density waves. b This shunt current represents the fraction of Imain for which the attendant magnetic …eld is cancelled by the reaction …eld due to the induced rotor current.11c). However the rotor loop impedance 0:5 X2 + R2 .main (primary magnetizing b inductance) and produce a large self-inductive voltage Emain.

5.b = sPgap.b ) (5.22) (5.3.f = sPgap.f + Protor.20) 5.3 Torque The electromagnetic torques of the forward and backward ‡ ux-density waves (…elds) are Pgap.f + (2 s) Pgap.5 Output and E¢ ciency The output mechanical power is Pmech = ! m Tmech = (1 s) ! s Tmech = (1 s) (Pgap. Derivation of these results is exactly analogous to the derivation of Tmech in Section 1.f Tmain.f and Protor. This occurs because the "backwards" power "opposes" forward power.9. !s (5.2 of the previous notes entitled Polyphase Induction Machines.5.19) because the torque of the backward …eld opposes that of the forward …eld. (5. Derivation of these results is exactly analogous to the derivation of Protor in Section 1.b ) . Finally.17) !s and Tmain.b = 1 (Pgap.f !s Pgap. This undesirable e¤ect is one reason that the performance of polyphase induction motors is better than that of single-phase induction motors. ANALYSIS BY EQUIVALENT CIRCUIT 109 stationary rotor (s = 1) and diminishes as s decreases.b = (2 s) Pgap.f Pgap.21) respectively.b (5.3.3.4 Losses The I 2 R losses due to forward and backward waves follow from the corresponding portions of the equivalent circuit in Fig.f Tmain.11c): Protor.9. slip (2 s) plays the role of slip s in derivation of Protor.b .18) respectively. Note that the interaction of the forward and backward waves will also produce an opposing torque of zero average value which ‡ uctuates at double frequency.f = (5. The net torque is Tmech = Tmain.b = Pgap.3.b . 5.1 of the previous notes entitled Polyphase Induction Machines. 9.b . the total rotor resistive power losses are additive because the rotor currents due to forward and backward …elds are at di¤erent frequencies (orthogonality): Protor = Protor.

f Pgap. . 5. quadrature) two-phase supply voltages. the current in the second winding must be phase-shifted with respect to the current in the main winding.5 5. Pgap Pgap. The e¢ ciency is (1 s) (Pgap.4 Split-Phase and Two-Phase Motors A second stator winding will create starting torque if the combined e¤ect of both stator windings is to produce a forward ‡ ux-density wave with greater magnitude than that of the backward ‡ ux-density wave. Refer to Fig..AND TWO-PHASE INDUCTION MOTORS where the last expression follows by substitution of equation 5..b ) : Pmech = = .1 of the textbook. Analytically a split-phase motor is equivalent to a two-phase motor operating under very unbalanced conditions. E. di¤erent resistance). The auxiliary winding is switched out automatically (e.1 Split-Phase Con…gurations General In a split-phase motor the second winding is called the auxiliary (or "start") winding. by a centrifugal switch) after startup.f + Pgap.19. In the former case it is called a split-phase motor because the required phase shift between winding currents must be obtained by a di¤erence in the phase angles of the complex impedances presented by the respective winding circuits.3 and Section 9. The analytic approach is the same for split-phase and two-phase motors: decomposition into forward and backward stator mmf waves.g. 5.g. The simplest method to obtain phase shift relative to the main winding is to increase the resistance of the auxiliary winding. Normally the distribution of the second winding is in spatial quadrature with that of the main winding. In the latter case it is called a two-phase motor (or symmetrical two-phase motor if two identical windings are in spatial quadrature). but the two windings may have di¤erent individual distributions and di¤erent electrical characteristics (e. the auxiliary winding may be wound with …ner wire which can tolerate the required operating current long enough to start the motor.. An induction motor with two stator windings (as described above) may be designed for connection to a single-phase voltage supply or designed for connection to a two-phase voltage supply.5.110 CHAPTER 5.2.b so it is ine¢ cient to operate the motor at large slip. To achieve the desired effect.g. 9. Ideally the AC currents in these windings are in quadrature (90 phase shift) but a smaller phase shift can work too. Phase shift in a two-phase motor results by connection to either unbalanced or balanced (equal. SINGLE.

3.2. The synchronous reluctance motor only starts as an induction motor while the hysteresis motor is not an induction motor. and 9. and Example 9.2 and Practice Problem 9. BALANCED TWO-PHASE SYSTEMS 111 5.5.1. and hysteresis motors.6. 5.4 Examples Refer to Example 9. at 90 orientation to each other. E.e.. 9.5.i.5 of the textbook. Permanent Split-Capacitor Motor The series combination of winding and capacitor is not switched out after start. Rather this combination is used to improve the operating characteristics of the motor at normal speed.g. Refer to Sections 9. 9.2.3 Other Other motor types are shaded-pole induction motors. This achieves better starting torque than the simple scheme above. Assume a reference coordinate system in which a positive angle corresponds to counterclockwise rotation and for de…niteness the b winding axis .2 Capacitor-Type Motors A capacitor is inserted in series the auxiliary winding to optimize the phase shift (typically to 90 ) between the main and auxiliary winding currents. self-starting synchronous reluctance motors. 9.5.6 Balanced Two-Phase Systems The sinusoidal two-phase currents ia and ib (two-phase voltages va and vb ) are called balanced only if they are equal in magnitude and 90 out of phase. Capacitor-Start Motor The series combination of winding and capacitor is switched out after start.6. These are not split-phase motors in the same sense as the above but share some similar principles.2. 9. One value optimizes starting torque and the other value is optimal for operation at normal speed.7. Refer to Fig..4. torque pulsations can be reduced Refer to Fig.2 of the textbook. This de…nition derives from the context of a system where the corresponding identical a and b stator windings are in spatial quadrature. Refer to Fig. respectively. except that the series capacitor is switched between two di¤erent values. Refer to Section 9. for additional details.5.2 5. 5. Fig.2.5.4. Capacitor-Start Capacitor-Run Motor This resembles the permanent split-capacitor motor. 9.

26) (5.2.28) and F + : Fa = = Fa + Fb : Analogous to Section 5.and backward.17 of the previous notes entitled Rotating Machines. and the cosine arguments ! e t and ! e t 90 are due to the time (phase) o¤sets of ia and ib respectively.30) where the cosine arguments ae and ae 90 are due to the spatial o¤sets of windings a and b.24) Ib = Im \ + 90 =) Fb = Fb + Fb .27.travelling waves: F = F+ + F : Clearly + + F + = Fa + Fb (5. For a postive sequence the superposition of the forward mmf waves reinforces to produce a large forward wave and the superposition of the backward mmf waves cancels to produce zero backward wave. ae ! e t)] (5. the individual forward-wave components are Fmax 2 [cos ( ae ! e t)] (5.23) (5. SINGLE. There are two cases for balanced currents (balanced voltages): ia leads ib by 90 or ia lags ib by 90 (va leads vb by 90 or va lags vb by 90 ): The phase in the former case (latter case) is called a positive sequence (negative sequence) because the resultant current sheet hence mmf wave will rotate in the forward (backward) direction. (5. The proof here is analogous to that for a balanced three-phase system in Section 1.31) . The net mmf is one wave F = Fa + Fb which can be decomposed into forward. Hence : F + = (Fmax ) [cos ( from equation 5.112 CHAPTER 5. respectively. (5.25) (5.AND TWO-PHASE INDUCTION MOTORS is oriented at +90 with respect to the a winding axis. The balanced (positive sequence) two-phase currents (in phasor notation) are: + Ia = Im \0 =) Fa = Fa + Fa . It su¢ ces to prove the above claim for a positive phase sequence since the result for a negative phase sequence will then follow by simply reversing the a and b labels. where Fa and Fb denote their corresponding mmf waves.1.29) and + : Fb = Fmax 2 [cos ( ae 90 [! e t 90 ])] = Fmax 2 [cos ( ae ! e t)] .27) (5.

At a given point on the advancing wave ae Fmax 2 [cos ( ae + ! e t)] (5. SYMMETRICAL COMPONENTS 113 Also analogous to Section 5.a pair with positive phase sequence and a pair with negative phase sequence. For a time interval 4t this gives Npoles 4 2 or a ! e 4t = 0 This last expression is the actual rate of angular advance and agrees with ! s : 4 a !e i: =h Npoles 4t 2 5.36) !s = Npoles Proof. We conclude that F = F + + F = F + consists of only a forward-travelling spatial mmf wave whose amplitude is twice that of the forward-travelling wave due to a single phase acting alone.28.1. These two balanced pairs or "components" are symmetrical in the sense that their .32) Fmax [cos ( ae 90 + [! e t 2 Fmax [cos ( ae + ! e t)] 2 90 ])] = Fmax 2 [cos ( ae + !e t 180 )] (5.7 Symmetrical Components The symmetrical component method is a decomposition of a pair of arbitrary two-phase currents ia and ib into the sum of two balanced current pairs.7.33) (5.34) ! e t = constant or Npoles 2 a ! e t = constant.5.2. Theorem 1 The forward-travelling wave rotates at the synchronous angular velocity 2 : !e : (5.35) by equation 5. the individual backward-wave components are : Fa = and Fb : = = Hence F =0 (5.

.114 CHAPTER 5. the decomposition implies b b b Ia = IF + IB (5.4. This method can be applied to voltages as well as currents. in which case the results will be identical in form.9 Unsymmetrical Two-Phase Induction Motor This general case is analyzed by application of the machine inductance matrix.AND TWO-PHASE INDUCTION MOTORS equivalent current sheets are revolving in opposite directions with respect to the horizontal axis. 9.14. j VB applied. Finally.37) (5.) In phasor form. Note that the symmetrical component . this produces a balanced pair of stator winding currents and only a forward mmf wave. Then the motor is analyzed again with only the b b balanced pair VB .40) b Ia b Ib (5. and 9.8 Application of Symmetrical Components Symmetrical two-phase motors can be analyzed by symmetrical components. 5. j Ib represents the negative-sequence balanced currents.41) 5.13. j If represents the positive-sequence balanced currents and Ib . In matrix form these equations are: " # " # b b Ia 1 1 IF = (5. I. (Note that a similar development is possible for three-phase currents. as b b follows. refer to Section 9.3. b IF b IB # = 1 2 1 1 j j " b Ia b Ib # : # (5.12. this produces another balanced pair of stator winding currents and only a backward mmf wave. j VF applied. Arbitrary applied voltages Va and Vb are decomposed into symmetrical b b components VF and VB : The motor is analyzed once with only the balanced b b pair VF . Refer to Figs. the response to the original input voltages is determined by superposition of individual responses due to the forward and backward waves. SINGLE.2 of the textbook. " Finally " by evaluation of the matrix inverse. simply substitute V for I in the above equations.39) b b j j Ib IB where the solution is " b IF b IB # = 1 j 1 j 1 b Ib = b b j IF + j IB .e. 9.38) and b b b b where If . Refer to Example 9.

. UNSYMMETRICAL TWO-PHASE INDUCTION MOTOR 115 method cannot be applied directly in the unsymmetrical case because balanced rotor current pairs do not follow from decomposition of applied voltages into balanced pairs. a split-phase induction motor is unsymmetrical.5. E.9. Refer to Examples 9.g.5. .4 and 9.

116 .

DC motor applications include automobile starter motors. Previously you received a link to a website where Class Notes 6 are posted for DC Machines by Prof. e.g. or separately excited. traction motors for subways and locomotives. and variable-speed electric drill motors. economy.Chapter 6 DC Machines 6.e. it stems in part from the various possible operating …eld con…gurations. 7. The stator is salient pole with a concentrated (…eld) winding.1 Introduction DC machines have some desirable features: 1. I. while the rotor is non-salient (cylindrical 117 . Kirtley (MIT). the …eld winding(s) of the machine can be shunt-. This wide variety of characteristics is possible under steady-state or dynamic operating conditions. 4. 7. compound-. 2.2 Machine Construction Refer to Fig. 5. DC motors are and will continue to be widely applied even though AC machines have made some inroads due to the development of solid-state drives. relatively simple drive systems.1 for a diagram of a DC machine. 3. Kirtley. versatility. 6. 6. ease of control (precise output)... etc.Fig. series-. wide range of motor speeds. wide variety of volt-ampere or torque-speed characteristics. Here we will refer to a few illustrations from those notes.

4. For analytic simplicity each section can be viewed as a single loop or turn which can be enlarged to any number of turns with no loss of generality. 4. DC MACHINES or "round") with a uniformly distributed (armature) winding. E. as explained later. as shown in Fig. and the return (outer) conductor in slot 7. Note that the speed voltage induced in any turn of the armature winding will have the same waveshape as the ‡ ux-density wave in view of the ‡ ux-cutting principle.118 CHAPTER 6. 6.. indicated by small circles near the perimeter of the rotor. where the connection to the rotating armature winding occurs by means of the commutator located on the rotor shaft. where the air gap is uniform and small. These currents sheets have odd symmetry about the vertical armature axis. This ‡ ux-density wave alternates between positive and negative (‡ at-top and ‡ at-bottom) extremes. The ‡ tops and bottoms (constant sections) of the wave correspond to the at faces of the salient poles. Loop conductors. run parallel to the long axis of the cylinder where currents entering and leaving the page are indicated by crosses and dots. respectively. DC current is supplied to the …eld and armature windings.3 Stator B Field The operating B or ‡ ux-density …eld in the air gap due to the concentrated stator (…eld) winding is by design an approximation to a stationary square wave as a function of the angle around the gap periphery. These gradual transition regions are desirable to reduce spurious e¤ects such as eddy currents. as shown in Fig. since the right-hand .18a and reproduced by the short-dash curve in Fig.1 Rotor Armature Winding The armature winding on a cylindrical rotor is shown schematically in Fig. The commutator functions as a mechanical switch or "recti…er" necessary to maintain a spatially …xed DC current distribution (sheet) in the rotor.17 for an illustration of a simple (two-segment) commutator in an elementary DC machine.4 6. Refer to Fig. the …rst loop consists of the outgoing (inner) conductor in slot 1. Refer to Fig. The sloped transition regions of the wave correspond to the large air gaps which occur outside the arcs of the pole faces and these transitions are by design gradual relative to those of a true square wave. 1. 7.7a for the speci…c case of a two-pole machine with 12 sections placed in corresponding rotor slots.consisting of an outgoing conductor and a return conductor spaced at 180 . In general each of the 12 loops are o¤set in sequence by 30 on the perimeter of the rotor but are otherwise identical. the dotted back-end connection from slot 1 to slot 7. 4. 7.16 for a cutaway view of a DC motor with multi-segment commutator in its right-hand side.11.4. The armature sections or loops are connected to establish a pair of DC current sheets on the perimeter of the rotor.g. 6.

Referring to Fig. the armature B …eld is directly proportional to the mmf wave in the region corresponding to the faces of the salient poles. The overall connection of loops is shown in the developed form of Kirtley. as shown in the developed view of Fig.2 Commutator The commutator is shown schematically in the center of the rotor in Fig. Refer to Fig. commutator segments 1 and 2 are attached to the outgoing and return conductors of the …rst loop.. Note that the axis of symmetry of the current sheets. As the rotor rotates. 2.7a. This mmf wave can be derived by application of Ampere’ law with integration of the current sheet (density) within the s appropriate magnetic paths through the stator and rotor cores.23 and reproduced by the dash curve in the developed view of Fig. where the air gap is uniform and small. as shown in Fig. Clearly the current in L(1) after transfer must be equal and opposite to that before transfer. 6.9.4. which is essentially a series connection where commutator segments are attached as periodic "taps". The numerous separate copper segments of the commutator are imbedded at uniform intervals in insulating material around the rotor shaft. 7. ROTOR 119 and left-hand current sheets enter and leave the page. The commutator is shown in developed form in Kirtley. 6.8 which illustrates "linear commutation".Fig.7 where the rotor rotates counterclockwise. The armature B …eld follows from the armature mmf wave with due regard for the variation in gap as a function of the angle around the gap periphery. The mmf corresponding to these current sheets is a stationary sawtooth wave as a function of the angle around the gap periphery.6. A network equivalent circuit for the armature in terms of lumped inductances is shown in Fig. respectively.. 4. the commutation process is essentially the transfer of one armature loop at a time from one current sheet to the other current sheet. Maximum armature mmf occurs on this axis. This occurs while loop 1 is brie‡ short-circuited as the commutator brush spans segments 1 and 2. However subsequently current must build in the opposite direction as well. I. respectively. which lies at 90 to the direct axis. is referred to as the quadrature axis of the machine. 7. On the other hand the armature B …eld is .Fig. 7. loop 1 is transferred from the right-hand to the left-hand current sheet while coincidentally the diametrically opposed loop 7 is transferred from the left-hand to the right-hand current sheet.4.4. Consider the transfer of loop 1 which is equivalent to the transfer of L(1) in Fig.g. where the black rectangles indicate commutator brush contacts for access from the external armature terminals. 7. so a transient voltage must appear across L(1) because di y v (1) = L(1) dt . 7. 7. during which period a normal exponential current decay occurs. so an additional transient voltage is required.3 Armature B Field The armature (B or ‡ ux-density) …eld in the air gap is due to the aforementioned pair of current sheets. E. 2.9. 7.e.

2) : Npoles Ca where Ka = 2 m : This follows from equation 4. 4.120 CHAPTER 6. 9 and Fig. See KirtleyFig. Based on the approximation that d originates from a spatial ‡ ux-density wave which consists of only a fundamental sinusoid. where Npoles Ca 2 m d !m (6.18b.4.5 Speed Voltage The net speed voltage which appears between a pair of diametrically opposed commutator segments is the sum of the individual loop speed voltages because these loops are essentially in series.. 7. Due to the commutator action. m = number of parallel paths through armature winding. 6.53 in the textbook where : : Ca = total number of conductors in armature winding. the commutated (recti…ed) speed voltage for a single loop is shown in Fig. the average value of the commutated (recti…ed) net speed voltage is Ea = or Ea = Ka d ! m (6.4 Interpoles The commutation process is susceptible to sparking or arcing due to the transient voltage required to reverse current in a loop. This is key to avoiding large transient voltages because it drives the reversal during the short-circuit condition rather than during the subsequent connection to the opposing current sheet. Interpole windings are an additional measure which e¤ectively reverses a loop current by creating an additional B …eld in the neutral zone where the stator …eld is zero. 6.2 In the case where the spatial ‡ ux-density is a ‡ at-top wave as in Fig. note that the individual loop speed voltages are also progressively shifted in phase due to the angular displacement of the individual loops around the rotor. This is illustrated in Fig. so the resulting induced voltage can drive the current reversal through the brush.21 for illustration of interpoles. Note that the average value Ea is the same whether the Ca armature conductors are concentrated or distributed. DC MACHINES greatly reduced in the region corresponding to the large air gaps which occur outside the arcs of the pole faces. E. and d is the direct-axis air-gap ‡ per ux pole.g. 4. However in large machines additional measures may be necessary. but the ripple is greatly reduced if the conductors are uniformly distributed.18b. 7.1) . This additional …eld is placed where a loop is brie‡ shorty circuited by the brush. the net speed voltage seen at the brushes approximates a constant (DC) waveform where the ripple is greatly reduced due to the addition of these phase-shifted speed voltages. In small machines adjustment of the brush position and linear commutation conditions can be su¢ cient to eliminate this problem.

4) from the above equation with @B = @B @ by the chain rule for di¤erenti@t @ @t ation. Here reduction of ripple is equivalent to reduction of harmonic content in the Fourier series.2 Eddy Currents Eddy currents will be induced in the rotor if it "sees" a changing magnetic …eld. This is illustrated in the developed diagram of Fig. a changing magnetic …eld B induces an electric …eld E according one of Maxwell’ equations (Faraday’ law in di¤erential form): s s O E= @B : @t (6. 6. 6. In the DC machine the resultant B …eld wave is stationary while the rotor turns through the wave. Also eddy currents may not be a serious problem if the rotor angular velocity ! is su¢ ciently small. In other words the magnitude of the induced electric …eld (hence eddy currents) is proportional to @B for the stationary resultant B …eld wave.6. This leads to i2 R losses so the core may be laminated to reduce the e¤ect.6. I.6. 7. 6.3) The electric …eld then causes circulating or eddy currents to ‡ due to the ow conductivity of the rotor core.1 Magnetic Field Analysis Resultant Field The resultant B …eld in the air gap is the sum of the …elds due to the stator (…eld) winding and the rotor (armature) winding. Coincidentally the rotor B …eld has inherently gradual transitions because it is a sawtooth wave. A direct sum is taken because both the stator and rotor …elds are basically radially directed in the gap.11. @ : given a speci…c mechanical rotor angular velocity ! = @ : Hence the stationary @t stator B …eld wave is designed with the aforementioned gradual transitions between its extreme ‡ at-top and ‡ at-bottom regions.3 Cross-Magnetizing Armature Reaction This form of armature reaction results from the interaction of the armature …eld and the magnetically nonlinear core material.e.6. However for multiple phase-shifted loops the commutated sum will have a waveform akin to that in Fig..2 where sine rather than ‡ at-top waves are used for illustration. Hence O E= @B @ @ @t (6.6.6 6. First consider the resultant . 7. MAGNETIC FIELD ANALYSIS 121 there is a substantial ripple due to the original transition regions in the stator B …eld.

This e¤ect can be viewed in terms of an equivalent reduction in the e¤ective mmf available from the …eld winding(s) hence an approximate translation of the magnetization curve for zero armature current. The e¤ect of cross-magnetizing armature reaction can be presented in a set of magnetization curves where armature current is a parameter.7 of previous notes entitled Rotating Machines.5 Torque Torque was derived from coenergy of the air-gap magnetic …eld in Sections 1. 7. 6.5. torque.18.18.122 CHAPTER 6.6. Here the …eld due to the armature mmf sums directly with that produced by the stator winding. 6. These windings mirror the armature ux windings and are connected in series with them so as to be "self adjusting" with respect to armature current. The latter statement follows because the armature portion of the …eld moves with the armature so produces no net e¤ect in Faraday’ law concerning rate of ‡ change in a conducting loop. Refer to Fig. Refer to Kirtley.Fig.14 for magnetization curves. Generalization yields a result which consists of the sum of many terms like the right-hand side of equation 6. where sr .11 for the ideal case where there is no armature reaction. The net e¤ect is that only the ‡ produced by the ux …eld winding remains. That derivation was based on the assumption that both the stator and rotor spatial ‡ ux-density waves are sinusoids of fundamental frequency and the most convenient form of result for present use is: T = Npoles 2 2 sr Fr 2 sin r: (6. to produce a uniform tilt in the ‡ top of the resulat tant ‡ ux-density wave.6 and 1. in other words the tilt in the ‡ top of at the resultant becomes non-uniform. However the net ‡ per pole does not change and the ux speed voltage does not change.a demagnetizing e¤ect. E¤ectively the armature mmf has acted to inhibit the ability of the core material to respond to the stator mmf. In this case the resultant ‡ ux-density wave is not the simple sum of the individual waves. s ux Cross-magnetizing armature reaction occurs because nonlinear saturation of the core material produces a non-uniform response to the additional mmf created by the armature. Fr .22 for illustrations of compensation windings. and ux speed voltage will change. Hence the net ‡ per pole.5) That derivation can be generalized to arbitrary stator and rotor spatial ‡ uxdensity waves by using Fourier series representations for both.4 Compensation Windings Compensation windings are wound in the …eld pole faces to cancel cross-magnetizing armature-reaction ‡ in large machines. and sin r in each term correspond to a speci…c harmonic in the Fourier series. 7.6. DC MACHINES B …eld shown in Fig. (It turns out that there will be no "cross-product" . 10 and Fig. 7.

2 and equation 6. This amounts to dropping the harmonic terms from the full generalized expression. This gives Pmech = Ea [Ka Ka d d ia ] = Ea ia .11) The mechanical power can be shown equal to the electromagnetic (gap) power Ea ia .10 into equation 6. In the present case equation 6.5 as an approximation for torque in the present case where the ‡ ux-density waves are not pure fundamental sinusoids. Hence T = Npoles Ca 2 m d ia (6. due to the orthogonality property.5 specializes to T = Npoles 2 2 d Fa1 2 (6.6) where sr = d is the direct-axis air-gap ‡ per pole and Fr = Fa1 is the ux fundamental component of the spatial armature mmf wave. by substitution of !m = Ea Ka d (6. m = number of parallel paths through armature winding. Fr .7) Fa1 = 2 where (Fag )peak = Ca 2mNpoles / ia (6. We rewrite the last equation as T = : where Ka = Npoles Ca 2 m : Ka d ia (6. and ia = armature current.8 into equation 6. POWER 123 terms between sr .12) from equation 6. Note that all even harmonic terms are zero anyway for a square wave or sawtooth wave.7 Power Pmech = ! m ( T ) (6.13) .7 followed by substitution into equation 6.9) by substitution of equation 6. (6.6.) Hence we can use equation 6.8) : : with Ca = total number of conductors in armature winding.7.11. and sin r for di¤erent harmonics.10) 6. Note that sin r 1 because the angle r between the armature and resultant mmf waves is nearly 90 : By Fourier series 8 (Fag )peak (6.6.

.13.5 and 7. as illustrated by Fig. will produce a variety of machine characteristics. E. 7.g.6 for generators and motors. These various connections.9 Field Winding Connections The …eld winding can be separately excited.7. 7. DC MACHINES where rotational losses must be subtracted to determine power at the machine shaft.124 CHAPTER 6. which is called a compound connection.3. . 6. 7. 7. etc. The total electrical power (excluding series and/or shunt …eld windings) at the machine terminals is P = Ea ia + i2 Ra a : where Ra = armature resistance. and 7. 7. 6.5. respectively. Refer to Section 7.6 for analysis of some generator and motor connections.2. connected in series.4 and 7. or connected in parallel with the armature.8 Examples Refer to Examples 7. illustrated in Figs. It may also consist of a parallel (shunt) and a series winding. for a series connection If = Ia . 7.1. 7. Analysis of machine characteristics for the various connections and parameter values begins by imposing the connection contraints on the machine variables.4.6.