Вы находитесь на странице: 1из 7

I E E E Transactions o n Energy Conversion, Vol. 14, No.

4, December 1999

982

On Equivalent Circuit Modeling for Synchronous Machines


Gordon R. Slemon and Mohamed L. Awad
Department of Electrical and Computer Engineering,
The University of Toronto,
Toronto. Ontario Canada M5S 3G4

Abstract: This paper presents alternative equivalent circuits


for modeling the SSFR frequency-response data measured for
turboalternator synchronous machines. These models provide
direct repreaentation of the frequency-dependent skin-effect
properties of the dampers and rotor iron. The parameters are
determined by mptching the measured data to the model frequency response using standard optimization methods. The
accuracy of matching is comparable t o that of conventional
third-order d-q models but the required number of parameters
is reduced.
Keywords: Synchronous machine; Modeling; Skin effect; Frequency response; Equivalent circuit.

I. INTRODUCTION

there is a direct linkage between a circuit parameter and a section of the machine, there is a basis for improved insight into
the explanation of the machine's behavior. Also, fruitful communication between system analysts and machine designers
is promoted when the models used in analysis and operation
can be related directly to those quantities under the control
of the designer. The approach of most system analysts seems
to have been to accept the properties of the machine. Little
attention has been devoted to communicating to the designer
the properties which might be more desirable.
It is the intention of this paper to present a critique of
the conventional equivalent circuits models for synchronous
machines and to suggest alternatives which may be useful in
system analysis. Following normal practice in such transient
circuit models, nonlinearity due to mametic saturation will
usually be ignored in this paper. This assumption allows the
conventional d-q axis approach with its provision for superpositian to be adopted. The authors intend to address the effects
of magnetic nonlinearity in a subsequent paper.
A corollary which follows from the assumption of superposition iS that the major space fields within the machine must be
assumed to be effectively sinusoidally-distributed in circumferential space around the air gap. This is not a necessary assumption in developing separate direct and quadrature circuits
bot is usually implied in superposing the direct and quadrature variables to obtain the terminal voltages and currents and
the developed torque.
~

The analytical modeling of large synchronous machines has


received extensive attention in the literature. Why then a further paper a n this topic? The direct reason for this present
work is that an improved basis was sought far application of
a concept developed for synchronous machine modeling which
is intermediate between conventional equivalent circuits and
finite element analysis [l], In this concept, subdivisions of a
one pole section of the machine are represented directly by
elements of an equivalent circuit. A considerable amount of
information is available on certain large synchronous generators from steady state frequency response ( S S F R ) tests and a
number of papers have been written on the derivation of direct
and quadrature axis equivalent circuit models which approximate these frequency responses [2]-[7].
The major limitation of these conventional models from our
point of view is that the structure and parameters of these
circuits are inadequately related to the physical dimensions
and the material properties of the synchronous machine. If
PE-129-EC-0-06-1998 A paper recommended and approved by the
IEEE Electric Machinery Committee of the IEEE Power Engineering
Society for publication in the IEEE Transactions on Energy
conversion. Manuscript submitted November 17, 1997; made
available for printing June 12, 1998.

11. A SIMPLIFICATION
OF CONVENTIONAL
MODELS
Fig. 1 shows conventional transient flux-linkage/current circuits for the direct and quadrature axes. These are third order
models, the most complex usually employed. In these models,
A d , A, and A, are the flux linkages of the stator direct axis,
the field and the quadrature axis respectively. LIis the leakage inductance associated with the stator winding, consisting
of slot, harmonic and end-winding components plus a portion
of the air gap circumferential leakage. L,r and L,, represent
mainly the linear air gap reluctances in the direct and quadrature axes. The quantity n, is the effective field/stator turns
ratio and is given by:

Nir
n, = -

N.,
in which N i e and N.. are the equivalent numbers of sinusoidally distributed turns in the field and in each stator phase
winding. In each case, the equivalent sinusoidally distributed
turns Ne are related to actual turns N by:

0885-8969/99/$10.00 0 1998 IEEE

983
4

Ne = - k d k , N
U

(2)

where k d and lip are the distribution and pitch factors for
the winding.

Fig. 1. Conventional models for synchronous machines


(a) Third-order direct axis circuit.
(b) Third-order quadrature axis circuit.

Circuit parameters in the field circuit can be referred to the


stttor reference frame by nse of the ratio n.. For example, the
actual field circuit resistance RIreferred to the stator becomes:

3
(3)
2%
where the 3/2 factor Links the 3-phase stator to the single phase
rotor.
The linear direct axis circuit of fig. 1-a can be replaced by
the simpler circuit of fig. 2 without any loss of information
relating terminal quantities. This is accomplished by use of
a simple transformation which was introduced for induction
machine analysis about three decades ago [E], [9]. The result is
now generally designated as a r-type circuit in contrast to the
T-type circuit of fig. 1 because of its structure at the stator
end [lo]. This transformation has been reinvented frequently
since that time in various forms. More recently, its validity for
use with synchronous machines has been established independently by Kirtley in [Ill.
Rfd

=Rf,

Fig. 2. Simplified r farm d-axis equivalent circuit


In this simple transformation, all the flux linkages to the
right of Lad in fig. 1-a are multiplied by a factor y d while all
currents are divided by y d where:
(4)

and

Ld

is the unsaturated direct axis inductance:

In the new r-type circuit, the leakage parameter L L D to the


right of L d is then given by:

All other inductances and resistances parameters in fig. 2 are


then made equal to those of fig. 1-a multiplied by f d 2 . The
new direct axis parameters are distinguished by use of the
subscript D rather than subscript d .
The equivalent transformation ratio nE for the new circuit
is:

An immediate advantage of the circuit of fig. 2 is that it


contains one fewer inductance parameter than does fig. 1 with
no loss of ability to predict terminal information.
There has been considerable discussion in the literature
about the nature of and need for the parameter L f l 2 d in fig.
I-a [3], [SI, [ 6 ] . Its presence in the conventional circuit has been
shown to be essential in predicting torsional behavior [12]. The
magnitude of this parameter is usually small but in some optimizations and for Borne machines this parameter takes on a
negative value. This is a result that may be mathematically
useful but physically not realistic.
It is suggested that an explanation may follow from the almost universal practice of fixing the stator leakage inductance
LI at a value obtained from the machine designer. This inductance certainly contains components arising from the stator
slot leakage, the stator end leakage and the flux due to stator
space harmonics. But it should be noted that the division of
leakage inductance between stator and rotor is somewhat arbitrary in view of the air gap circumferential flux component
which is usually significant in large gap turboalternators.
The unsaturated inductance Loa is physically and primarily
identified with the reluctance of the air gap. If all or most
of the circumferential leakage is included in the stator leakage
inductance L I , a negative value far L f m may well be required
on the rotor side to produce good match to SSFR test results.
In the simplified circuit of fig. 2 all of the leakage components
up t o the surface of the rotor have been combined into the parameter L L D . With magnetic linearity assumed, the division
of leakage between stator and rotor is irrelevant in analyzing
machine behavior and the anomaly of potentially negative inductance is avoided.
If parametem of a r-type circuit of fig. 2 have been evaluated, the conventional circuit of fig. I-a can be produced if
considered necessary. Given a value for LI from the designer,
the inductance L a d is found from ( 5 ) , y d from (4) and the inductance L f l l d from ( 6 ) . All D-subscripts parameters to the
right of L f l m can be converted to their d-subscript values by
dividing by yd'. The effective turns ratio n. can be determined
by use of (7).
An additional advantage of the circuit of fig. 2 is that the
inductance L d and the ratio n.z can be obtained directly from
steady state open-circuit and short-circuit tests without any
required input from design information. If, at field current if,
the rms short circuit current is I, and the unsaturated open
circuit voltage (i.e. on the air gap line) is V., then:

984
Ld

V,
W J .

and:
3
--

I d

(9)

A if

The ratio ne in the conventional circuit is a well defined


quantity easily computed from knowledge of the numbers of
slator and rotor turns, their distribution and their pitch. It
may be argued that, if any quantity is to be requested from
the manufacturer, it might preferably be this ratio. Given this
quantity, the appropriate value of the factor T d can be determined by comparison with the ratio nB obtained as in (9) and
used to obtain appropriate values of Lt and L a d .
The usual reason for insisting an the retention of the conventional form of circuit is that magnetic saturation in the m e
chine is commonly msociated with the parameter L a d . Assuming that the permeability of the iron sections of the machine
is high, this inductance represents primarily the reluctance of
the air gap of the machine and therefore is inherently linear.
Magnetic saturation occurs, not in the air gap, but partly in
the stator teeth and yoke due to the stator flux linkage and
partly in the rotor teeth and core due to the rotor flux linkage.
Associating saturation with the parameter L a d is therefore a
compromise. This issue will be addressed in a subsequent p"
per dealing with saturation effects.

111. C O M M E N T A R Y O N THEP A R A M E T E R S

OF

CONVENTIONAL
CIRCUITS
The currently preferred method of deriving equivalent circuits is by parameter fitting to a set of results obtained by
standstill frequency respmse tests [3]-[7]. The usual tests give:

Much of the discussions associated with models has been


directed at the accuracy with which the models match the
S S F R test data over the measured frequency range. Little attention is given to specifying the accuracy and the frequency
range that is to be considered adequate for various types of
studies. As an example, the highest frequency considered to
be significant for torsional vibration studies in 60-Hz turboalternators is considered to be about 55 Hz.In stability studies,
the frequency range of greatest interest is usually considered
to be 0.1 to 5 Hz 171.
Ideally, the models used far a study should be the simplest
ones that gives adequate accuracy for that type of study. In
the following sections, an alternate approach to modeling will
be developed, one based on L physical assessment of the rotor
structure in an attempt to produce simple models which are
adequate to predict performance. Primary attention will be
given to a consistent modeling of L d ( 8 ) and L , ( s ) since the
interaction of the stator with the attached system is of greatest
importance in most stability, hunting and resonance studies.

IV. AN ALTERNATEAPPROACH
SSFR test results have been made available t o the authors
for a 500-MW Ontario Hydro turboalternator designated as
Nanticoke-5. Some significant data on this machine is listed
in appendix A. A sketch of a 90-degree cross-section of the
rotor is shown in fig. 3. This rotor has uniformly distributed
aluminum damper bars in both direct and quadrature axes, the
3 bars near the direct axis being somewhat smaller than the
6 bars near the quadrature axis. The slots below the smaller
bars are filled with a non-conducting material.

d axis

damper bars.

L d ( s ) , the direct-axis operational inductance being the


ratio of flux linkage ,id to i d with the field shorted,

L q ( s ) ,the ratio of A, to i,,


sG(s), the ratio of

if to i, with field shorted, and

L . f o ( s ) , the ratio of

to

id

with field open.

All of these are expressed as Laplace transforms, usually


graphed for magnitude and phase over a. frequency range which
may extend from
to 10' Hertz. Model structures are
chosen for direct and quadrature axes such as the third order
circuit of fig. 1, or the second order models which result from
deleting L 2 d , R w , L 8 , and Rs, from fig. 1. The parameters
are iterated mathematically to obtain a best fit typically using
a least squares method. In developing a direct axis model,
greatest weight is u s u d y assigned t o the function L d ( S ) . In
general, the fitting results improve as the order and complexity
of the model is increased. In Borne studies, fourth or higher
models have been tried in order to obtain I more accurate
fit [I3].
The shunt branches in the conventional circuits such as those
shown in fig. 1 represent the best compromise of linear parameters available for the order chosen. It is not appropriate to
associate any one branch with the physical damper winding in
that axis.

-field

slots

Fig. 3. A 90-degree cross-section of the rotor

A . Steady State Parameters


The following parameters were obtained from open and
short circuit test data listed in the appendix:

L d = -15430
6572

and :

- 2.35 pu

985

RE

this minimum angle


Then,

3 6572

-= 13.93
.A 1001

OCCUTS.

For this machine j m , n = 0 . 5 j IIr.

"-

The measured function L d ( s ) is shown in fig. 4. The parameter L d may be derived from the measured magnitude ( L d (
as frequency approaches zero. This value of Ld = 2.15 pu
is somewhat lower than that obtained from the open/short
circuit tests in (10). This is usually attributed to the IOW permeability of the stator and rotor iron at the low values of flux
density used in S S F R tests. Finite element analysis has shown
that this relative permeability can be as low as 150 for S S F R
tests. Considering that the flux path length in iron is about
30 times the gap length and that the iron permeability for
the apen/short circuit tests would be an order of magnitude
higher, the difference of about 9% between the two values of
L d is readily explained.

Lp
RF

tan [ 4 5 = 1.12 s
(13)
2nfmcn
where R F is the field circuit resistance. The base angular
frequency w b is 371 rad./s. then the equivalent field resistance
R f d is:
j-,7=-=

Rfd =

LF
= 0.00105 pu
TFWb

(14)

This may be compared with the value calculated from the


field resistance R j using the measured current ratio R E to
transfer it to the stator reference turns. From (3):

Rfd =

3 4 = 0.00102 pu
2REzZ,b

(15)

It is seen from fig. 6 that this first order model for L d ( S ) is


reasonably accurate up to about 0.1 Hs. For higher frequency,
the effects of damper and rotor iron conductivities must be
included.

'-r'.q"y.nry,Hzi-

Fig. 4. Variation of L d ( s ) with frequency.

B. First Order Model


In the direct axis, the field winding dominates the transient
behavior particularly at low frequency. Let ns first consider
the extent to which the simple first order model shown in fig.
5 can represent the measured function L d shown in fig. 4.

v
L

Fig. 5. First order direct-axis model


If a first order model applies, the value of the field leakage
inductance L F can be derived by noting the minimum value of
the angle adm." of & ( s ) which is 45' in fig, 4 far this machine.
Given the magnitude of the angle;

from which L F = =~ 0.445 pu. The field time constants


may now be estimated by noting the frequency fmin at which

Fig. 6. Frequency plots of first order model for Ld(8).

C. Damper Modeling
A model is required that accommodates various configurations of damper bars. In the Nanticoke-5 machine, the duraluminum damper bars extend over the whole of the rotor
periphery as shown in fig. 3. The bars centered about the
direct axis have about 0.75 the cross-sectional area of those
around the quadrature axis.
At low frequency, the current distribution in each bar can
be considered uniform. Noting that the fields for each axis
can be considered as sinusoidally distributed about that axis,
the effective law-frequency damper resistances R ~ and
D RbQ
reflected into the stator reference frame can be determined by
weighting the resistance of each bar by the value of the sine
function appropriate to its bar position. These resistances will
also include provision for the end winding resistance.
With excitation from the stator, as frequency increases the
current in each bar is forced toward the surface of the bar near
the air gap. The effective penetration or skin depth in the bar
is:

986
6a = pod)-^'^
(16)
where u b is the conductivity ofthe bar material and f is the frequency. Far the duraluminum allay of the bars in Nanticoke5 , with conductivity of about 2 107mho/m,this penetration
depth is approximately 'I5 mm. With a bar depth of about 40

J;

mm, the current distribution can be considered to be approximately uniform far frequencies where the bar depth is less than
36, i.e. for frequency less than about 1 Hz. With skin effect,
the damper bars of depth d can be modeled approximately in
each of the axes as au impedance of the form [14]:
lib

Za = Ra-

sinh(ka) + sin(ka)
cosh(kn) - COS(kb)

(1- j) ohm

(17)

where ka = 2 d / 6 b . For flux linkage models, the corresponding


operational inductance is:

The damper model requires only two parameters, Rao and


in the direct axis, RbQ and k a in
~ the quadrature axis. In
the Nanticoke-5 machine, the depth of the bars is the same in
both axes so that only a total of three parameters are needed.
In this modeling, the variations in the width of each bar are
ignored.
kbD

D. Iron

An approximate model for the iron can be expressed in


terms of a single constant li, which lumps together the results
of (20) and (21), the proportion of the rotor periphery in iron
and transfered to the stator reference frame:

li.
Lci(3) = --(I

Jif)

-3)

with separate values of k , for


~ the direct axis and Iraq for the
quadrature axis.

E. Model Structure
The proposed structure far direct and quadrature axis models is shown in fig. 7. With stator excitation at high values of
frequency, the currents in both the damper bars and the rotor
iron are farced to near the rotor surface. Since these currents
are side by side around the periphery, their model elements in
fig, I are properly connected in p a r d e l . In these r type circuits, the parameters L L D and LLQ include the stator leakage
and all of the air gap circumferential leakage up to the rotor
surface. In the direct axis, the parameter L F D includes the
leakage around the field winding plus the leakage across the
damper bars.

Modeling

In the iron of the rotor, high frequency currents will be


forced to near the rotor surface. At lower frequencies, the iron
current may extend down the teeth and into the rotor core.
In order to obtain a reasonably simple model, let us ignore
the variations with depth in the width of the available current
paths and use a simple approach based on the skin depth,

6;= ( ~ ~ , p o u ; f ) m
-~.~
(19)
where p, is the relative permeability of the iron and U ; is the
iron conductivity.
Far the cast steel of the rotor, the conductivity is about 3
10' mho/m. The relative permeability of the steel is estimated
to be about 200 for the low values of flux density that apply
during S S F R tests. These values result in a penetration depth
is approximately
mm. This suggests a penetration depth

Ji
tooth equal to the depth of the damper

into I rotor
bars at
a frequency of about 0.15 Ha and a penetration depth to the
bottom of the quadrature axis rotor slots at a frequency of
about 0. 01 Hz. From this it may be concluded that skin
effect is a major factor in representing the iron over most of the
frequency range of interest in stability and resonance studies.
A square surface of material with conduction limited in
depth by skin effect can be cansidered to have an impedance
of:
Z(s) =

(-)"'"

As an element in
ductance:

(1 j) ohms/square

Fig. 7. Developed model for synchronous machines


(a) Direct axis circuit.
(b) Quadrature axis circuit.
As frequency is reduced, the mean levels of the induced
currents in the damper bars and the iron will be increasingly
below the rotor surface. It may be argued that the appropriate
point of connection of the damper and iron elements should
be moved somewhat toward the field terminals as frequency
is reduced. For simplicity, this refinement will be ignored.
However, this dependence of structure on frequency provides
insight into the reason for the ladder type structure which
characterizes some higher order circuits such as that shown in
fig. 1.

F. Eualuotion of Parameters
(20)

flux linkage model, this becomes an in-

An optimization routine was used t o evaluate parameters


for the proposed circuits to match best the measured S S F R
results for the functions Ld(s) and L q ( a ) . The inductances L d

987
and L , were taken from the asymptotic values as frequency
approached zero leaving 6 parameters to be evaluated for the
direct axis and 4 for the quadrature axis.
Each optimization was based on minimizing the sum of the
squares of the magnitudes of the differences between the measured and model vectors divided by the magnitude of the measured vector as in the general expression:

This optimization gives equal weight to a given percentage


error thus allowing far the considerable variation in the magnitude of the function with frequency.
The results of this optimiiation are shown in fig. 8 and fig.
9. It is observed that the accuracy of the models is quite good
particularly over the upper frequency part of the spectrum.
For the L d ( s ) function, the largest magnitude error is about
2.7% at 0.14 Hz and the largest angle error is about 1.4' at
0.08 Hz. In the frequency range 0.5-50 HE of major significance
for stability and resonance studies the error is generally below
0.8% in magnitude and 0.6' in angle, For the L g ( a ) function,
the largest magnitude error is about 3% at 0.13 HE and the
largest angle error is about 1.8' at 0.35 Is.

The parameters obtained from the optimizations are listed


in table 1. It is noted that the value of the field resistance
R F Dis approximately equal to the values derived earlier in
(15). The sum of the inductance parameters L L D and L F D
is approximately equal to the value 0.445 derived for the first
order model in section B. The damper resistances R D and
Rg are approximately equal to the values calculated from the
dimensions of the bars and end rings. The constants k for
the two axes differ by an amount which is consistent with the
differences in bar and iron tooth dimensions.
TABLE I

MACHINEPARAMETERS

[ Parameters I

pu value

G. Transfer Function sG(s)

Fig. 8. Frequency plots of &(a)

using developed model.

Fig.9.Frequency plots of L q ( s ) using developed model.

In this paper the focus has been on modeling the machine


looking into the stator as this is the function most needed for
stability and torsion studies. The measured function sG(s)
was not used in the optimization of parameters. As a matter
of interest, the d-axis model of fig. 7 with the parameters of
table 1 were used to predict the function aG(8). The result
is compared with measured values in fig. 10. It is noted that
the errors in magnitude and angle are considerably larger than
those of L d ( s ) and L g ( s )but the general trend may be good
enough to be useful.

Fig. 10. Frequency plots of sG(s) using developed model.

988
A reason for the discrepancy in modeling this sG(s) function
is that the skin effect models used assume that high frequency
current is forced to near the rotor surface. With high frequency
currents in the field there will also be skin effect in the iron
around the w d s of the field slots and there may also be skin
effect at the bottoms of the damper bars. A model with an
additional skin effect element connected at an inductance point
closer to the field could be considered to accommodate this
effect.

V. CONCLUSIONS
An alternate set of equivalent circuit models has been presented to represent the frequency response of a turbogenerator
as Seen from its stator terminals. Skin effect elements dependent on frequency are used instead of the constant parameters
of conventional models.
The accuracy of modeling is comparable to that obtained
with third order conventional circuits. The number of parameters to be evaluated in the optimization is 6 for the direct axis
and 4 for the quadrature axis. These may be compared with 7
for each of the direct and quadrature axe8 in the conventional
third order models.
Each parameter of the models can be evaluated directly
from the dimensions and material properties of the generator
and thus provide a direct communication link to the machine
designer.The models are directly applicable for frequency response analysis. They are not readily applicable where time
constants are required.
The authors welcome comments and suggestions from the
user community on the concepts which have been introduced
for discussion in this paper.

APPENDIX
Data far Ontario Hydro Generator - Nanticoke-5.
Rated MVA: s b = 588.2 MVA
Rated stator voltage: V b = 22 kV rms line-to-line
Rated stator current: I b = 15.347 kA
Base stator impedance: Zab = 0.823 ohms
Base stator i n d u c t a n w l s = 2.183 mH
Stator resistance: R. = 1.579 m ohm (25'C)
Field resistance: Rt = 0.106 ohm (25OC)
Field turns: N I = 120
Stator turns per phase: N . =7
Stator winding pitch 7/9
Field current for rated stator voltage on air gap line = 1001 A
Short circuit stator current with lOOlA field current = 6572 A

ACKNOWLEDGMENT
The authors express their thanks to their colleagues Prof. P.
L. Dandeno and Prof. M. R. Iravani for their valuable advice
and comments and to Geoffrey Klempner of Ontario Hydro
for making available the SSFR data and the physical and
magnetic properties of the machine. The authors also wish to

thank the Natural Sciences and Engineering Research Council


for research funding.

REFERENCES
Slemon, "An equivalent circuit approach to analysis of
synchronous machines with saliency and saturation," I E E E
Paana., Vol. E C - 5 , N o . 8, pages 538-545, 1990
[2] IEEE Committee Report (P.L.Dandeno Chair), "Supplementary definition and associated test methods for obtaining parameters for synchronousmachine stability study simulations,"
IEEE Tmns., Vol. PAS-39, No. 4, pages 1625-1633, 1980
[3] LM. Canay, "Determination of model parameters of machines
from the reactance operators "&a), zp(ja): I F E E Trans.,
Vol. EC-8, No. I, pages 272-279, June 1993.
[4] I.M. Canay, '"Modeling of alternating-current machine having
multiple rotor circuit," IEEE Trona., Vol. E C - 8 , N o . 2, pages
280-296,June 1993.
[5] I. Icamwa, and P. Viarouge, '"Onequivalent circuit structure
for empirical modeling of turbine generator," IEEE Il.ons.,
Vol. E&$, No. 8, pag& 579-592, Sept. 1994.
[a] S. Homing, A. Kehani, and I. Kamwa, "On-line evaluation
of round rotor synchronous machine parameter set estimated
from standstill time-domain data," Pa-149-Ee-0-01-1997
[7] "IEEE test procedure for synchronous machines", IEEE Std
115, 1995.
[8] C.R. Slemon, "Magnetoelectric devices-transducem. transformemand machines," John Wiley and Sons,New York, 1996.
[SI M.Poloujadof, and Ivenes, "Comparison des achema equivalents au moteur asynchrone ployphase," Rwre Generals de
I'Electricite Vol. 76, No. 1, pages 1-6, Jan. 1967.
[lo] G.R. Slemon, "Modelling of induction machines for electric
drives," IEEE T~ane., Vol. IA-25, No. 6 pages 1126-1131,
Nov/Dec 1989.
(111 J.L.Kirtley, "On turbine-generator mtoi equivalent circuits,"
I E E E Trans., Val. PWRS-89. No. I , pages 262-271, Feb. 1994.
1121 V. Atarod, P.L. Dandeno, and M.R. Iravani, "Impact of synchronous machine constantti and models on the analysis of torsional dynamics," I E E E Trona., Vol. PWRS-87, No. 4 , pages
1456-1463, Nov. 1992.
[13] A. Keyhani, and H. Tsai, "Identification of high order synchronous generator from SSFR test data," IEEE Trans, Vol.
EC-9, NO.'^, pages 593-603, Sept. 1994.
[14] S A . Swenn, and J.W. Salmon, "Effective resistance and reactance of a rectangular conductor placed in a. semi-closed slot,"
IEE Proc., VoI.110, No. 9, 1666-1662, Sept. 1963.
[l] G.R.

Gordon E.Slemon is professor emeritus in Electrical and Computer Engineering at the University of Toronto where he served as
Head of the Electrical Engineering Department from 1966 to 1976,
and as Dean of its Faculty of Applied Science and Engineeringfrom
1979 to 1986. Currently he acts as a consultant to industry and
government as well as continuing his research specialty of electric
machines and drives. He is the author or coauthor of five textbooks
and 160 technical papers. He established the Innovations Foundation at the University of Toronto and is a co-founder and director of
the company, Inverpower Controls Ltd. He is President-Elect of the
Canadain Academy of Engineering. an Honorary Fellow of the Institution of Electrical Engineers (UK) and ai Fellow of the Inetituteof
Electrical and Electronics Engineers and the Engineering Institute
of Canada. In 1990, he received the IEEE Nikola Tesla Award and
Gold Medal.
Mohamed L. Awed received his B.Sc. and MSc. degree from
Cairo University, Egypt, all in electrical engineering in 1990 and
1993 respectively. Since 1990 he has been with the Electrical Engineering Department, Cairo University. He is currently a Ph.D. candidate at the University of Toronto. His research interests include
power system simulation, dynamicsand control, and AI applications
in power systems.

Вам также может понравиться