IEEE Transactions on Power Systems, Vol. PWRS-2, No.

4, November 1987 833

ADMITTANCE MATRIX MODEL OF A SYNCHRONOUS MACHINE

FOR HARMONIC ANALYSIS
A. Semlyen J.F. Eggleston J. Arrillaga
University of Toronto University of Canterbury
Toronto, Ontario, Canada Christchurch, New Zealand

Abstract - The harmonic analysis of a power system requires
appropriate models of all system components. Synchronous
machines act as harmonic converters, sensitive to the sequence of
the fundamental and harmonic frequencies. This paper describes
the derivation of a harmonic model of the machine in the form of a
three-phase complex admittance matrix and its application to the
harmonic behaviour of an asymmetrically loaded generator.
INTRODUCTION
The importance of harmonic analysis of power systems is on
the rise [1] due to an increased use of converters which are the pri-
mary source of harmonics. Several papers have been published on
this topic, some describing the problem as harmonic power flow
[21,[3], some as harmonic penetration [4], and a recent book deals
with power system harmonics [5]. In addition to converters, also
arc furnaces, fluorescent lamps and the magnetizing branches of
transformers [6] are recognized sources of harmonics. However, the
harmonic behaviour of the synchronous generator has not been
given serious consideration in harmonic analysis 14], even though it
is well known that it converts negative sequence currents into third
harmonic positive sequence and, in general, acts as a harmonic con-
verter. This is so probably because of the complexity of the prob-
lem, since conversion means coupling of harmonics which otherwise
would be examined separately. Therefore, generators are often
represented by a single approximate impedance at each harmonic.
However, three-phase transmission lines appear as strongly unbal-
anced at harmonic frequencies and resonances may appear for single
modes so that harmonic unbalance is created or strongly amplified
[7]. The generator will pick up the unbalance and return other har-
monics. Clearly, the complex interaction betweein generator and
system cannot be ignored.
analysis are superimposed on the base load flow solution and
represent an increment to it. The harmonic analysis will contain all
significant frequencies, including d.c. and the fundamental fre-
quency. The reference of the latter serves also as a phase reference
for the harmonics. Since the harmonic analysis is superimposed on
the base load flow, the generator will be assumed as having a
shortcircuited field winding, which makes the harmonic model com-
pletely passive. Equation (1) corresponds to this condition.
HARMONIC MODEL OF A SYNCHRONOUS MACHINE
Derivation of the Matrix Ydqh
The synchronous machine admittance model is derived from
the d , q -axis differential equations. Generally accepted conventions
and notations are used, as detailed for instance in Ref. [8] and illus-
trated in Fig. 1. With two damper windings in the rotor, s and t,
the equations are:
Vd Rid + P (Ld id + Mdf i + Md, i, w(Lq iq + Mqt it)
vq = Riq + p (Lq iq + M it ) +w(Ld id + Mdf if + Mdi
vf =R i + p (L i + Mdf id + Mf8 i,) 0°
(2")
vs = Rs i, + p (L8 it + Mdq id + M0. if ) 0
vt = Rt it + P (Lt it + Mqt iq ) = O
In (2) vf has been set to zero because, as mentioned, the field
voltage has been considered in the base load flow. However, the
parameters of the excitation circuit have to be included in Rf and
Lf All variables are phasors of harmonic h:
I

This paper describes the derivation of a synchronous machine

model and its application to the analysis of an asymmetrically wh= Re {WI e hwt } = W'hcos (hwt) W" h sin (h wt ) (3)
loaded synchronous generator. It is a three-phase model because of
the significance of negative sequence in harmonic conversion. It coIn- where
tains all harmonics (even and odd, as these turn out to be uncou-
pled) and permits the calculation of the current vector Wh Wkh
W + jW" h (4)
i = Yabc V (1) denotes any phasor. Bold is used to denote a complex number
and/or a matrix or vector. We have now
Data for synchronous machines are normally available in
terms of d-q axis and implicit in this information is the effect of p = j Cw (5)
the second harmonic terms of the inductances. Information on
fourth harmonic terms, absent from the d -q axis model, is not gen- and equations (2) become algebraic operations in Vd,, Vq,, ldh, Iqh I
erally available and is ignored in this investigation.
It is also assumed that a base load flow solution has been
If,h,1h It.. The last three can be eliminated from (2'), using (2")
so that we obtain
obtained for the power system, balanced out at the fundamental
frequency, so that a standard load flow program can be used. The q
harmonic analysis is performed subsequently on the unbalanced sys-
tem in a three-phase representation. The results of the harmoniic

S6 SM 35i0-3 A paper recomnended and approved

by the IEEE 'Power System Engineering Committee of
tl'e IEFE 'Power Engineering Society for presentation
at the IEEE/PES 1986 Suinmer Meeting, Mexico City,
d
Ilexico, -July 20 - 25, 1936. Manuscript submitted
.January 18, 1985; made avrailable for printing
May 5, 1986.
Printed in the U.S.. Figure 1 Basic machine representation

0885-8950/87/1100-0833$01.00© 1987 IEEE

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
 834
and
Vdh Zlh Z2A ]1
V dh
(6) 2 ( dA + V'q1) V- h
1 = (V dh -F Vq")
Vor Jn Z3m Z4c fqor
2 1hh 1
I~(VId (12))
(11)
v' h 2 (V" dh + V' q,) v" + V"q
or, in more compact form

Vdqh - Zdq idqh (7') These can be combined into phasors, as in (4), yielding:

and ah-, 2 (dh I Vqh )

idqh Ydq Vdqb (7") (1 3')

-
ah+, 2 (Vdh + Vqh )
For simplicity, we will denote the elements of the non- and
symmetric matrices Zeh or Ydjq by

ah bh1 V4- 1 - 2 (j Vd4 + gh )

Ch dh
(8) (13")
h +j 2=-( iVdh + Vqx )
A particular matrix Zdh of interest is for the rotor freque!ncy
h =0 . Then p becomes zero and equations (2') yield In equations (11), (12) and (13) the superscript h indicates the
related d ,q -axis harmonic for a component of order h ±1 of a sta-
Z dq. =
R
wLd
-ZL[
R
] (8')
tor (a,#-axis) harmonic. When two such components of the same
order (related to a lower and a higher order rotor harmonic) com-
bine, the superscript is dropped; see the vector (16').
Equations (13) should be written for all h. However if h 0=
Der'ivation of Connection Matrix C or h 1 = 0, then VO - VO; and if h 0O, then we obtain
Although our aim is to derive an admittance matrix in phase directly from (9'), converted into phasor relations,
quantities, the mathematical formulation is simplified by using an
intermediate connection matrix in terms of a,f3 quantities, intro- Va,t Vd, + jvQlo (14')
duced by the rotation
VPi, =-jVd + Vq. (141")
va,,.l [Cos wt -sin wt Vd1
[VIj - sinwt wt cos v j (9') obtain
By assembling equations (13) and taking (14) into account, we

where h' denotes the resultant harmonics. The inverse transforma- va - CVdq (15')
tion is
- [-cswt
An identical transformation applies for currents:
Vd sinwt VCA

Vq ~ -sin wt Cos wt V oh (9") ia/i Cldq (15"1)

In (15')
Identical transformation applies for currents. If we write, as in
(3), ya2.va2,... Vctn
VaQp = | V,I Vl1, Vi,, # 9 V 1rT (16')
VdA - V' dACOS (h wt )- V dhsin (h wt ) (10')
vqh - V qh cos (h wt ) V" qh sin (h wt ) (10") Vd0-[ Vd, Vq., Vdl, Vl VVd2, Vq2, *.. ,Vd, V
qT] (16")
and substitute (10) into (9'), we obtain expressions for v,,, and vjh,
in the form of (3), with h' -- h ± 1: and
a == V' sh cos(h -1)wt - V" , sin(h - 1)wt + N
-t V, h +lcos(h + 1)wt V- ,h,sin(h + 1)wt (11') 2M N
M N
M N
h 1)wt +
V -cos(h -1)wt
Vup, '- - h sin(h
A - C- 1 (17')
2
+ v'f,hcos(h + 1)wt - VI Ph sin(h + 1)At (11")
M N
where M N
V) M
VI (thh -I 2(VAdAdi, +V
2 h-l 2-(V"
(12')
V,+ - 2 Adh V qX ) r" - + V' I) with

Ij 1
(1711)
(17"--)

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
 835
where the asterisk denotes the conjugate and n the highest har- where
monic investigated.
The matrix C is singular because it contains pairs of rows (or AO = NYdqN / 2
columns) with matrices M or N only, which are singular due to the A1 - (2MYdq, M + NYdq2N) / 4
fixed relation between the a and /5, or the d and q components,
respectively. This singularity can be removed by deleting the first A =
(MYdq,_,M + NYdlh,+N) / 4 (for A - 2,...n -1)
(or the second) row and column of the matrix and by leting its size (23)
increase indefinitely (or, alternatively, deleting also the last row and
column). In this way, only one of VU, V#, and of Vd4,, VY will B'1 = MYdqIN/2
remain in each of the vectors vr and vdq of (16) (and only one of B h -= MYdq,,N / 4
the last two elements in these vectors if the truncated matrix C is (for h -2,...,n -1)
considered). Keeping only one of V,,,, VP. does not result in loss B'^ = NYdqAM/4 (for h 1.n- 1)
of information since VP - jV,. , Similarly, it can be seen from
(17') that the sum Vd. +jVq. itself affects the value of Val and of
An ZMYdqg,_M
-
/4
VP rather than its individual components Vd, and Vq}. Any of the
latter, having assigned an appropriate complex value, can replace These expressions can be calculated from:
the sum Vd, +IjVq.
MYdq, M _((ai +di ) - j (b, -s ) )M
Let C' denote the reduced matrix C, and let C' and C* be
their conjugate, respectively. Then it can be easily verified, by NYdqN -((ai+di) + j(bi -c))N
direct substitution and taking (17") into account, that C'C'* -U,
where U is the unity matrix. This does not mean however that C MYdq, N ( (ai -di ) + i(bi ci )P (24)
is the inverse of C. Still, we prefer to use the full matrix C of (17')
because of its symmetry and simplicity, with the understanding NYdqM- ((a-d, ) -j(bjci) Q
that a reference to an inverse relationship to (15) is interpreted to
apply to the reduced matrix C' and the reduced set of variables. where
The inverse relationship to (15) can be derived directly starting
with equations (9") and performing calculations as in equations (10)
to (14). We obtain

Vdq - C . Xa (18X)
P -= [_ -1 ' Q -

From equations (23) we can see that the self admittance AA

1 (25)

(181) for the harmonic

of order h depends only on the rotor
la# of order h d1, and the mutuals B'A and B' ' between frequencies
idq =
harmonics
h+1 and h -1 depend on the rotor frequency of order h. These
Derivation of Matrix Y.# results are consistent with the well known pertinent physical
phenomena. An intuitive block diagram of the voltage and current
We assemble equations (7"), for all h, into the relation: harmonic relationships in the armature and rotor is shown in Figure
2.
iiq -Ydq Vdq (19) The matrix Y,,,# has one non-zero block band on each side of
where Ydq is block diagonal with blocks of the form (8). the diagonal and a zero block separating band. Therefore even and
odd harmonics do not interact in the generator.
For given vj we obtain Vdq from (18') then id from (19) and
The impedance matrix Z,,,f can be obtained similarly to (21):
i,# from (15"). Alternatively, we can write
iap1Y3.1fie (20) Za,"'- CZdq C' (26)
where

Y:-CYdq C' (21)

The non-symmetric matrix Yafj has the form I \\

AO B,
B' '
1

' '
Figure 2 Harmonic interaction 1.e,tween generator and system
A1 B" Derivation of Matrix Yabc
B'I A2
B' t h We first augment Y,1, by a zero sequence diagonal matrix,
Y - B'2 (22) comprising all harmonics:
Ah
B, '-

Y
ta/oo
=

YojJ (27)
B'n -1

It is generally assumed that zero sequence currents produce no

significant resultant flux in the air-gap and accordingly equation
(27) shows no coupling between the zero sequence and the com- a,/5

ponents. At least one author [9J disv u('s briefly such an effect, but
stresses that it is negligible and subsequently removes the coupling
from the analysis.

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
836
For any harmonic we can write From equation (32)
Vapol, - Th Vabch o-
o -Ygens v + i (35)
(28)
iaflo# -= Th iab,h
However, when the excitation is perfectly smooth, i, is only
where affected by positive sequence fundamental frequency current. There-
fore equation (35) becomes
1 1 1
2 2 io -Ygenv + i+ (36)
Th = 0
2 2
(29) where vr and i+ are vectors in the phase frame of reference contain-
1 1 1 ing only the positive sequence fundamental components of v and i
-3 3 3 respectively. While v+ is kept fixed at 1 p.u., i+ is not known until
the final solution. However i+ can be expressed as a function of i
using
Equations (27), (28) and (29) can be completed and assembled
to yield the following: if T i (37)
a/Jo Tv abc where
(30)
i T-11 flo Fundamental Harmonics
1abc =T a2
1 a
An admittance matrix a2 1 a 0 Fundamental
a a2 1
T=
1 (37')
Yabc YT-0Y(o T (31) 3

could be calculated, but the sequence (30) is equivalent to the 0 0 Harmonics

intended use in (1) (where the subscripts abc have been omitted
from v and i).
APPLICATION OF THE MODEL
TO AN ASYMMETRICALLY LOADED GENERATOR
The test system of Figure 3 is used to illustrate the effect of Combining equations (32), (34), (36) and (37) yields the following
the harmonic admittance model of the generator. It consists of a expression for the generator current
single circuit line of flat configuration corresponding to the Islington
to Kikiwa 220 kV transmission line in the New Zealand system [71. i= -YgenZcI +Ygenv + Ti (38)
The transformer is Yg /Y. connected, and its rating is 100 MVA,
14kV/220kV, with a leakage reactance of 0.112 p.u. or

Equation (1) is modified to include the effect of the fundamen- Ai b (39)

tal power flow solution. This is achieved by adding a positive
sequence fundamental current injection i, which maintains a con-
stant (1 p.u.) positive sequence component of fundamental voltage
where
at the generator terminals. Taking into account the sign convention
indicated in Figure 3, the generator current becomes A=U+ YgenZ9 - T

i -Ygenv + io (32) and

Because there is no coupling between odd and even harmonics,
vectors i and v and the generator admittance matrix Ygen contain
only the odd harmonics of Yabc-
un*ansposed Ike
Figure 3 Test system
The input impedance matrix of the system (Z, ), as seen from
the generator bus, can be derived from existing programs 141, [71, so
that for each harmonic the following relationship applies
VAh ZSi ih
b Ygen +
Equation (39) is solved for i by Gaussian elimination and vh is
obtained from (33) for each harmonic.
The transmission line introduces resonance and standing wave
Iinnm effects which somewhat obscure the understanding of the generator
harmonic contribution. Thus the problem is analyzed in three
stages, i.e. with the generator connected to (i) an unbalanced resis-
tive dummy load, (ii) an unbalanced tuned load and (iii) an untran-
sposed transmission line via a transformer. Details of the generator
used in the test example are given in the Appendix.
Unbalanced Resist'ive Load
The load is represented by the following admittance matrix
(33)

and for all harmonics (40)

Yo +- ==
Y+ KY1
v- Zi (34) KY+ Y+
where Z. is a block diagonal matrix.

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
 837
where
Y0 is a zero sequence conductance
Y+ is a positive (and negative) sequence conductance
K is a coupling coefficient between the positive and nega-
tive sequences
Since the generator produces no zero sequence voltage, Y0 can (a) (b)
be chosen arbitrarily; to prevent a singularity its value is made Figure 5 Simplified equivalent circuit of the machine
equal to Y+ in our example. Thus the three-phase admittance
matrix becomes (a) direct axis
(b) quadrature axis
6
1+ K K
2K

3 3 3 Voltage (%)

5
Y.bc = Y+ K
3
K
1- K 2K
33 (41)
-K 2K K
3 3
1-3 3 4
L

This matrix, although not physically realizable, can be used to

13
examine the effect of generator loading and degree of unbalance.
Voltage (%)
10 2
8
6 -3

4
2 25.0 50.0 75.0 100
saliency (%)
0
C 1.1
I

0.2
coupling coefficient
Figure 4 Variation of positive sequence third harmonic voltage
with coupling and loading
Figure 4 shows the effect of coupling and loading on the level
of 3rd harmonic voltage distortion. The results indicate that for
loads within the nominal rating (i.e. 1 p.u.) the 3rd harmonic vol-
tage is almost directly proportional to the load admittance, and
hence to current or power. Also, the 3rd harmonic voltage is seen to
be directly proportional to the level of coupling. The 5th harmonic
voltage, not shown in the figure, varies approximately in proportion
to K2.
Effect of Saliency
It is apparent from the analysis that saliency is the main
determining factor in the process of harmonic conversion. To exam-
Figure B Effect of saliency on generator harmonic voltages
The effect of saliency has been tested on a purely resistive
load of 484[ (1 p.u.) with a coupling coefficient of 10%. Cases of
perfect saliency (Mqj = 0) and zero saliency (M5j = 1) produced
no noticeable effect with perfect coupling (i.e. zero leakages). How-
ever, the addition of some leakage (0.2 p.u.) showed considerable
difference in the resulting levels of harmonic voltage distortion, as
showni in Figure 6.
Unbalanced Tuned Load
This case introduces the effect of resonance by replacing the
resistive load with the delta connected circuit of Figure 7, the reso-
nance frequencies approximating those of the open circuited line of
the test system. The machine data are given in the Appendix.
The sequence components admittance matrix of the delta cir-
cuit is

ine the effect of saliency, the simplified equivalent circuits in Figure

5 are used for the machine's direct and quadrature axes respec- 0 0
tively. These circuits include only one winding in each axis of the Yo+-
rotor (i.e. Md, = Mf, -0). Further simplification is achieved by
= O (3 + 6) Y -baY, ( 13)
setting the direct axis magnetizing inductance Mdf to 1 p.u. and O -Sa 2y 3 + 61
making Ld = L1 , Lq = Lt and Rf = Rt and assuming equal
leakage in both axes (i.e. Ld Mdf = Lq Mqt ). Finally saliency,
- -

defined as the ratio of quadrature to direct axis rotor fluxes, can be

expressed by 7 B

S =I1-M (42)
Mdf
Thus from specified leakage, saliency and resistances R and Rf,
the machine parameters of equations (2) can be calculated.
While this model is not representative of an actual machine, it c
does provide a simple way of varying and displaying the effect of
saliency on voltage distortion. C

Figure 7 Tuned delta load

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
 838
where
a = 1 [4200
Y2 == (1 + 6)Y1, valid at a particular frequency
and, using the previous definition of coupling coefficient,
3K-I
|3 +- 6 |(4
The values of L1, C1, L 2 and C2 are chosen to satisfy three
conditions, namely, to give the same positive sequence admittance
at the fundamental frequency, to produce specified resonance fre-
quencies f o 1 and f o 2 for Y1 and Y2 respectively, and to maintain
a specified degree of unbalance (determined by the coupling
coefficient) at fundamental frequency. The three resistances are
assumed equal.
From the test system with a line length of 200 km, approxi-
mate values are derived for the delta branch parameters; at 50 Hz
the reactance (mostly capacitive) and resistance are 1555 and 12
ohms, respectively.
In a real transmission line of flat construction the asymmetri-
cal input impedance gives rise to two line modes which resonate at
slightly different frequencies. This effect is simulated in the dummy
load by maintaining a centre resonance frequency f 0 and varying
the actual resonant frequencies of Y1 and Y2 symmetrically on
both sides of fo. Figure 8 shows the variation of harmonic voltages
with f o 1-f 0 2 while keeping f , constant at the third harmonic.
By maintaining a constant unbalance, the amount of negative
sequence voltage and current at fundamental frequency is approxi-
mately constant with varying f , 1- f o 2-
With t oi > f o 2 Y1 is capacitive and Y2 inductive and vice
versa with f o 2 > f o l. This results in a strong unbalance at 3rd
Voltage (p.u.)
harmonic, leading to increasing negative sequence 3rd harmonic vol-
tage and current with f o l-f o 2 This is followed by corresponding
increases of 5th harmonic voltage and current. The ratio of 5th har-
monic positive sequence to 3rd harmonic negative sequence is
approximately constant.
Untransposed Open Circuit Line
In the test system of Figure 3 the length of the untransposed
line was varied from 50 to 800 km, again with the machine dat of
(44) the Appendix.
Figures 9a and b show the positive and negative sequence
third harmonic voltages at the machine terminals. While the
machine cannot generate harmonics directly, these appear as a
result of the unbalance produced by the untransposed line and are
therefore very dependent on the length of line.
In the range of line lengths between 165 to 200 km each har-
monic voltage shows a double peak. These peaks correspond to the
different resonant lengths of the a and /3 propagation modes present
in a line of flat construction. The peaks occur at different lengths
for the different harmonics (e.g. at 170 and 195 km for positive
sequence third and at 175 and 190 km for negative sequence third).
This effect is due to the impedances and degree of unbalance vary-
ing greatly near the resonances of the two modes.
Voltage (%)
(a) negative sequence

-3
4.3 l7 line length (km)
1.0
6

4
(b) positive sequence

,.-l 3

I I I

100 200 300 400 500 600 700 800

50 line length (kim)
fif-if0
01 02
) (Hz) Figure 9 Variation of third haLrmonic voltage with transmission
line length
Figure 8 Harmonic voltage distortion versus f ol1-f o 2 with centre (a) negative sequence
resonant frequency f 0 = 150 Hz (b) positive sequence

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.
 839
The harmonic voltage levels in this range of distances are too [31 D. Xia and G.T. Heydt, "Harmonic Power Flow Studies,
large to be ignored, viz. 6w % of positive sequence third and 4% of Parts I and II", ibidem, Vol. PAS-101, No. 6, June 1982, pp.
negative sequence third harmonics respectively. This clearly demon- 1257-1270.
strates the need for detailed generator representation as proposed in [4] T.J. Densem, P.S. Bodger, and J. Arrillaga, "Three Phase
this paper. The levels of distortion are also substantially higher Transmission System Modeling for Harmonic Penetration
than the levels calculated in the dummy loads, as a consequence of Studies", ibidem, Vol. PAS-103, No. 2, Feb. 1984, pp. 310-
the standing wave effect of the line. 317.
CONCLUSIONS [5] J. Arrillaga, D.A. Bradley, and P.S. Bodger, "Power System
Harmonies", book, John Wiley, to appear in 1985.
A generalized steady state model of the synchronous machine
has been developed which can take into account any asymmetry or [61 H.W. Dommel, A. Yan, and Shi Wei, "Harmonics from
distortion present in the armature voltages. It has been shown that Transformer Saturation", IEEE Paper No. 85 SM 381-9,
when the field voltage is perfect d.c. the harmonic model of the presented at the 1985 IEEE Summer Power Meeting, Van-
machine becomes a passive admittance matrix. The level of couver, B.C.
interharmonic coupling has proved to be partly load related (i.e. (7] J. Arrillaga, T.J. Densem, and B.J. Harker, "Zero Sequence
affected by the size of load current and degree of asymmetry) and Harmonic Current Generation in Transmission Lines Con-
partly generator related (affected by saliency). Such harmonic cou- nected to Large Converter Plant", IEEE Trans. on Power
pling cannot be detected with present harmonic models, where the Apparatus and Systems, Vol. PAS-102, No. 7, July 1983, pp.
generator is short-circuited behind the suhtransient reactance. 2357-2363.
Computer results, with the machine connected to a dummy [81 D. O'Kelly and S. Simmons, "Introductionj to Generalized
(asymmetrical) load and to an untransposed transmission line, have Machine Theory", book, McGraw-Hill, London, 1968.
been obtained to corroborate the theory. They indicate that the
harmonics generated by the machine may often exceed the levels [91 C.V. Jones, "The Unified Theory of Electrical Machines",
prescribed by harmonic legislation and they need to be assessed book, Butterworths, London, 1967.
accurately. The effect of two different resonant modes have been [101 H.H. Hwang, "Unbalanced Operation of Three-Phase
demonstrated, leading to a strong unbalance and thus high voltage Machines with Damper Circuits", IEEE Trans. on Power
distortion. The main harmonic contributions from the generator are Apparatus and Systems, Vol. PAS-88, No. 11, November
positive and negative sequence third harmonic currents, which 1969, pp. 1585-1593.
therefore cannot be eliminated by generator or transformer connec-
tions.
APPENDIX
ACKNOWLEDGEMENTS
Generator Data
The authors would like to express their appreciation to the
Natural Sciences and Research Council of Canada and to the New
Zealand Energy Research and Development Committee for financial The generator data, based on that of Hwang [10], is as follows:
support.
Nominal power rating = 100 MVA
REFERENCES Nominal voltage = 14 kV
Nominal frequency = 50 Hz
[1] IEEE Working Group on Power System Harmonics, "Power
System Harmonics: An Overview", IEEE Trans. on Power R =0.005 p.u. Mdf = 1.0 p.u.
Apparatus and Systems, Vol. PAS-102, No. 8, Aug. 1983, pp. Rf 0.0005 p.u.
= Md, 1.0 p.u.
=
2455-2460. R, = R -0.02 p.u. Mqj 0.6 p.u.
=
[2] W. Song, G.T. Heydt, and W.M. Grady, "The Integration of Ld = 1.2 p.u. Mf, -1.0 p.u.
HVDC Subsystems into the Harmonic Power Flow Algo- Lg = 0.8 p.u. Ro = Lo -o00
rithm", ibidem, Vol. PAS-103, No. 8, Aug. 1984, pp. 1953- Lf = 1.2 p.u.
1961. Ls -1.0 p.u.
Li = 0.831 p.u.

Authorized licensed use limited to: RWTH AACHEN. Downloaded on June 29, 2009 at 05:35 from IEEE Xplore. Restrictions apply.