Вы находитесь на странице: 1из 8

# Digital simulation of a synchronous generator in

direct-phase quantities
P. Subramaniam, B.E., M.Tech., and O. P. Malik, M.E., Ph.D., D.I.C., C.Eng., M.I.E.E.

## Indexing terms: Synchronous generators, Modelling

Abstract
The paper describes a mathematical model for the simulation of a 3-phase synchronous machine using
direct-phase quantities, thus obviating the need for any transformation. Numerical solution using a digital
computer has also been described, and compared with digital simulation in transformed d-q-O- and
a-/3-0-axes models of a synchronous machine. The proposed model in direct-phase quantities enables a
unified approach to be adopted in the study of both symmetrical and asymmetrical conditions. Since the
constraints to be imposed are direct operating conditions, asymmetrical operating conditions can be studied
very easily. Modifications required in the model to simulate various types of faults are described. Versatility
of the proposed model is illustrated by the study of a single-line-earth fault with single-phase opening and
automatic reclosure. It is shown that this type of fault can be studied as simply as, say, a 3-phase fault.

List of symbols once the speed variation is taken into account, making it
p = operator d/dt imperative to resort to a computer solution if exact analysis
M = mutual inductance is required.
L = self inductance With the advent of modern computers, numerical methods
Lmd = ^/-axis magnetising inductance can be employed efficiently for solving nonlinear differential
Lmq = </-axis magnetising inductance equations. In such a case, a direct 3-phase model can be used
Ld = d-a\\s synchronous inductance for a more exact study of synchronous-machine performance.
Lq — (/-axis synchronous inductance With direct representation, various fault conditions, such as
Lo — zero-sequence inductance symmetric, unsymmetric and sequential conditions, can be
R — resistance simulated with ease and a solution obtained.
L(lQ, La2, Ms0 = inductance coefficients of the armature This paper describes a mathematical model of a synchro-
Maq = mutual inductance between armature and nous machine in direct-phase quantities and the procedure
<7-axis damper winding for simulation using numerical techniques. The differential
if0 = initial field current equations have been solved using the 4th-order Runge-Kutta
Em — peak value of armature phase voltage method. This method requires the evaluation of the right-hand
(ojMafif0) side of the differential equations four times in each step.
S = load angle with reference to the infinite To study the effect of harmonics in synchronous machines,
busbar considered negative for generator Dunfield and Barton6 have proposed a model using phase
action quantities. The computational procedure for exact calcula-
co — synchronous speed tions requires inversion of a 6 x 6 matrix and multiplication
v = instantaneous speed (o> — p8) of matrices at least four times during each step. The mani-
H = inertia constant, kWs/kVA pulation of the performance equations and the algorithm for
their solution presented in this paper requires the inversion
Subscripts of two matrices, and other computations to be performed
a armature only once at the end of a step. This procedure has been found
/ field to give sufficiently accurate results. Thus it makes a more
kd direct-axis damper winding effective use of computer time than the model described in
kq quadrature-axis damper winding Reference 6.
e transmission line This paper also describes the modifications required in the
basic model for the simulation and study of various kinds
of faults. The usefulness and versatility of the proposed
1 Introduction model is demonstrated by an illustrative example studying
Methods of predetermining the transient performance a single-line-earth fault with subsequent opening and reclosure
of synchronous machines have, in the past, made use of a of the faulted phase. This type of fault cannot be studied by
number of approximations, in view of the fact that a closed- any of the models in d-q-0 or a-jS-0 quantities without
form solution of the performance equations was laborious going back to phase quantities.
and in some cases was not feasible.1 To overcome some of
the difficulties, various transformations, such as d-q-O, oc-fi-O
etc., were introduced. The d-q-zxes model yields differential 2 Mathematical model
equations with constant coefficients. They are linear provided Fig. 1 shows the idealised7 synchronous machine with
that speed is assumed to be constant, which rules out logical two damper coils. Mathematical representation using d-q-0
and rigorous analysis of power swings.2 In addition, study and a-jS-0 quantities is given in Appendix 9. In 3-phase
of unsymmetrical faults necessitates further transformation quantities, the performance of a synchronous generator con-
of d-q-0 equations. 3 - 4 In such cases, the a-/3-0 model has nected to an infinite busbar through a short transmission
been found to be more convenient to use.4 The a-jS-0 model line (see Fig. 2) can be described by equations given below.
results in differential equations with variable coefficients,
and an approximate solution has been suggested.4-5 As 2.1 Voltage relationships
stated earlier, the performance equations become nonlinear The position of the rotor at any instant is specified
Paper 6348 P, first received 4th August and in revised form 15th October with reference to the axis of phase a by the angle 6 (see
1970 Fig. i). In terms of flux linkages, voltage relationships for the
Mr. Subramaniam and Dr. Malik are with the Department of Electrical
Engineering, University of Calgary, Calgary, Alta., Canada. Mr. six stator and rotor circuits are
Subramaniam is on leave from the SV University College of Engineering,
e = px\t - Rl (1)
Tirupati, India
PROC. IEE, Vol. 118, No. 1, JANUARY 1971 153
7 P4
where e = [ea eb ec ef ekd ekq]> 2.3 Torque
Owing to inertia, the instanstaneous electric torque
differs from the prime-mover torque if the speed varies. At
R = diagonal [RaRaRa - Rf - Rkd - Rkq]

## When the machine is supplying a load through a short trans-

mission line (see Fig. 2), the machine-terminal voltages

da.
Fig. 2
System representation
d0
dt any instant, the mechanical torque is equal to the electrical
torque plus the accelerating torque. If Pm is the mechanical
power input, the mechanical torque is given by

T = —P
J r
m m
and the equation of motion can be expressed as

## q.o. As shown in Appendix 9.3, in 3-phase quantities, the electrical

Fig. 1 torque is given by
Schematic representation of a synchronous machine
T = {Mb - Q + *Pb(ic - O + M'a - i'b)}
e
cn eb-> ec c a n be expressed in terms of the infinite-busbar 3V3
voltage Emi as follows: (6)
«a = - Emi sin ojt + i(,Re + Lepia 2.4 General
Eqns. 1-3 and eqn. 5 form the complete model for a
= ~ E
mi sin
Lepib synchronous machine in phase quantities. The elements of
(2) the L matrix (expr. 4) are dependent on rotor position, which
varies with time. Hence the performance equations are
e
c = -
sin
+ y ) + 'cRe + differential equations with variable coefficients. If the speed
of the generator is varying under transient operating condi-
where cot = 6 + 8 and Emi is assumed to be equal to Em, tions, the performance equations become nonlinear. Generally,
the peak value of armature phase voltage at no load. solutions in closed form cannot be obtained, and numerical
solutions have to be resorted to.
The flux-linkage relationships of a synchronous gene-
rator can be expressed in symbolic form as 3 Digital simulation
<\>=LI (3) The simulation involves the solution, by numerical
methods, of the mathematical model developed above. Two
8
where L is
a b c

a -La0 — La2 cos 20 —Ms0 — La2 cos (20 — 2T7/3) -Ms0 - La2 cos (20 + 277/3)

## b -MsQ-La2 cos (20-277/3) -LaQ - La2 cos (20 + 2T7/3) -M,o-Lfl2cos(20)

c -Ms0 - L a 2 cos (20 + 2T7/3) —Ms0 — La2 cos 20 -La0-La2 cos (20 -277/3)

kd kq

## MO/COS(0-2TT/3) Maf cos (0 - 2T7/3) - M a ( ? sin (0 -- 277/3)

M fl/ cos (0 + 277/3) Maf cos (0 + 2T7/3) -Maq sin (0 + 277/3) (4)

£// Mfkd 0

Mfkd Lkkd 0 kd

"kkq kq
154 PROC. IEE, Vol. 118, No. 1, JANUARY 1971
of the most commonly used techniques are the Runge-Kutta Kutta method. This procedure has been found to give
method and predictor-corrector methods. sufficiently accurate results and reduces the computer time
The Runge-Kutta method, as modified by Gill, was used to one-third of that required for the exact analysis.
in the present investigation for the following rea sons: (viii) The currents and their derivatives computed above
are utilised to calculate the machine-terminal voltages given
(a) No special starting procedure is required, since the
by eqn. 2, and the flux-linkage-derivative vector is used to
method is self starting.
start the (n + l)th step of computation.
(b) It permits easy change of step length.
(ix) The calculations are thus advanced in single-step fashion.
(c) A straightforward computational procedure is repeated
During an interval, all variables are assumed to be constant.
throughout the calculation.
The additional computation required to determine the
(d) No modification of the computation is necessary for
current derivatives is necessitated since the digital model of
nonlinear equations.
a generator connected to an infinite busbar through a trans-
(e) The method is generally more accurate than other methods.9
mission line, as encountered in practice, is being considered.
From eqns. 1 and 2, the voltage equations can be rewritten This will not be required for simulating a machine directly
in matrix notation as connected to an infinite busbar. The accuracy of the overall
results depends considerably on the accuracy of matrix
p\> = e + RI (7) inversion which has to be performed at each step.
The basic algorithm for the simulation of the 3-phase model A simplified flow chart is shown in Fig. 3.
is as follows:
(i) The initial voltages, currents, current derivatives, and
rotor position, as specified by the problem, can be used to 4 Fault simulation
start the computation. For various faults occurring at the terminals of a
(ii) The flux-linkage derivative vector can be computed at loaded machine, the constraints to be imposed, and the
the beginning of the nth step with the knowledge of e and /. corresponding modifications required in the mathematical
(iii) On integration, this gives theflux-linkagevector *\>. model developed in Section 2, are described below. The
(iv) The electrical angle 6 at the nth step, which defines necessary changes to simulate faults or system abnormalities
the rotor position at that instant, can be obtained by solving can be incorporated at any stage of a study by a few simple
eqn. 5 for 8: instructions in the program. It thus allows any kind of study
(v) The angle d computed at the end of a step is made use to be performed uninterrupted. For the sake of clarity, only
of in forming the inductance matrix L (eqn. 4). those expressions or matrices that require to be modified are
(vi) The currents at the end of «th step are obtained from given.
the equation
/ = £-'d> (8) 4.1 3-phase fault
(vii) The current-derivative vector pi is obtained from the For a 3-phase-earth fault, at the instant of the fault,
following matrix equation, making use of the currents com- ea = eb = ec = 0
puted above:
l i
This will necessitate the following modifications:
p i = L ~ V + L ~ R m I —L ~ * p L I . . . . ( 9 )
N
M rrl tfl • »« tit * '

## The derivation of the above equation is given in Appendix 9.4.

The steps (v)-(vii) are performed only at the end of a step,
and not four times as required by the rigorous Runge- Rm — R
A 3-phase fault can be studied in a similar way. 10

## 4.2 Phase-phase fault

read initial For a fault on phases b and c,
conditions
e
bc = eb — ec — °
—| calculate py , p
Hence
calculate ^ • 8 and v e
b = ec

## form L matrix Also, ea + eb + ec — 0 since eQ — 0

calculate L
So, eb = ec = — -y at the instant of fault.
calculate
I = L "' * This condition is simulated with the indicated changes in the
following elements:
form modified V2 = V3 = \Emi sin wt
L m matrix
R
_ _ e _ _
R R
calculate m2, 1 - ^m3, 1 ~ ~ ~j m2,2 ~ ^m3,3 ~ a
p i vector
Lm2x=Lxx+hl Lml 2=L22

imposition of
conditions for
transient study

calculate the
armature voltage 4.3 2-phase-earth fault
for next step
For a fault on phases' « and b, ea = eb = 0. The
modified elements are then
Vx = V 2 = 0
= R R
Fig. 3 ^wl.l m2,2 = a
Flow chart Lm\, i ~ ^ i , i Lm22 = L22
PROC. IEE, Vol. 118, No. 1, JANUARY 1971 155
4.4 Single-phase-earth fault length, actual results are not included in this paper. However,
For a fault on, say, phase a, ea is equal to zero. This the results obtained from the two models were in complete
will necessitate the following modifications: agreement in each case, thus establishing the validity of the
3-phase model.

1,1

## 4.5 Open conductor

When circuits are controlled by fuses or a device that
does not open all three phases simultaneously, one or two
phases of the circuit may open on a fault while other phases
are still closed. A similar situation is likely to occur with a
broken conductor.
Neglecting the resistance of the arc which forms at the
time of opening, this type of fault can be studied by assuming -8
0 008 016 0-24 0-32
that the opening takes place at the instant when the current
time, s
in that phase is passing through zero in normal cycle. The
rate of change of current in that phase can be assumed to be Fig. 4
zero just after the fault. Thus at the instant when, say, phase Variation of torque for a 3-phase fault
a is open and thereafter, ia — 0. Just after the occurrence
of the fault, pia = 0. IO2r
The modified matrices are obtained by deleting the row and
column corresponding to phase a. The rest of the procedure
for computing the variables remains the same. The open-phase
machine voltage ea, which is the voltage induced by the mutual
10
fluxes, is computed at every step, making use of the mutual
inductances of phase a with other circuits, currents in other
circuits, and their derivatives.
fault on
Arc resistance can be taken into account by modifying the
resistance of the faulted phase during the period between the 098 L

occurrence of the fault and the instant at which ia goes 008 016 0-24 0 32
through zero. Assuming that the arc is quenched at this time , s
instant, the procedure indicated already can be employed. Fig. 5
This will have the effect of accelerating the decay of current Speed variation for a 3-phase fault
ia, and thus current in phase a will reach zero a very short
interval earlier than when arc resistance is not taken into 2 0
6
account and when current ia is assumed to reach zero in 6.
the normal cycle. However, the effect of this refinement on n phase open
the overall results will be insignificant.

## 4.6 Opening and automatic reclosure

Opening and automatic reclosure of the faulted phase
can also be studied using the procedure indicated in
Section 4.5. fault on reclosure
-20

(i)
5 Model verification and results
5.1 System studied
A simple system containing a synchronous machine
feeding into an infinite busbar through a short transmission
line has been studied (see Fig. 2). Details of the machine
and the system are given in Appendix 9.6.
Various faults applied at the terminals of the synchronous
machine have been studied. Variation of speed was taken
into account in each study. The generator currents, transient
torque and speed variation were computed in all cases.
-4
5.2 Comparison of the three models
5.2.1 System output
To verify the reliability of the 3-phase model, a sample
study of a symmetrical 3-phase-earth fault at the machine
terminals was made using the three detailed models described
in Section 2 and Appendix 9. Results obtained from all the
three models were in full agreement.
Figs. 4 and 5 show the plots of short-circuit torque and
speed variation with time. The mechanical input is assumed
to be constant during the fault. It will be observed from
Fig. 5 that there is a drop in speed immediately after the
fault, owing to transient torques developed by the interaction
of the fluxes and currents. Later, the speed increases as the
fault current decays. The consequence of this is a temporary 0-32
drop in load angle, referred to as 'back swing'. This result is
in agreement with that obtained in References 11 and 12.
Since the d-q-0 model is not well suited for the study of Fig. 6
unsymmetrical faults, such faults have been studied with the Computed armature currents for a single-phase fault with reclosure
3-phase model and the <x-/3-0 model. To maintain reasonable (i) Phase a (ii) Phase b (iii) Phase c
156 PROC. 1EE, Vol. 118, No. 1, JANUARY 1971
5.2.2 Computation time light of the limitations of the d-q-0 model, which is suitable
It has been observed that, to study unbalanced faults, for the study of symmetrical conditions only.
computing time for the 3-phase and a-/3-0 models is almost
the same, because the periodic nature of current, voltage 5.2.3 Computer-core requirements
and flux linkages necessitates a smaller step length. However,
The approximate total core requirements for the three
for balanced faults, a greater step length can be used for
models are as follows:
comparable accuracy, thus reducing the computation time in
approximately the same proportion as the increase in step d-q-0 model 4950 bytes
length. For the d-q-0 model, a larger step length can be used, a-^S-0 model 6120. bytes
in view of the aperiodic nature of the currents and voltages a-b-c model 6660 bytes
£ | (see Appendix 9.1) is not required at every step. It may be pertinent to point out that it is only the 3-phase
The time required for computing the results for a study model that is the most versatile. If the d-q-0 and a-jS-0
duration of 0-66s for a symmetrical 3-phase fault, using the
d-q-0 model, was 2min with a step length of 0-001s on an 10-02 xlO-'r
IBM 360/50 computer. For the other two models, the time
required for the same study duration was 6min with a step
length of 0-00035 s. It should, however, be looked at in the
•o 10

9-99
Z 4 0 008 016 0-24 0-32
time, s
Fig. 9

## II Speed variation for single-fault with reclosure

32
(0

28

w 0

24
008 016 0-24 0-32

-8 time , s
Fig.10
Variation of load angle for single-phase fault with reclosure
8r
models were to be modified to study all cases that the
3-phase model is capable of, their core-memory requirements
would be much larger than those indicated above.

## 5.3 Single-phase opening and reclosure

To study a single-phase-earth fault with single-phase
opening and automatic reclosure, the d-q-0 and a-^S-0
models in the form described in Appendix 9 cannot be
-8 employed. To apply the proper constraints, some of the
008 016 0-24 0-32
quantities have to be transformed into phase quantities; thus
time, s the advantages accrued through transformation are lost.
(iii) Using the 3-phase model described in Section 2, such a
Fig. 7
case can be studied just as easily as any other type of sym-
Computed rotor currents for single-phase fault with reclosure metrical or unsymmetrical fault. To illustrate the versatility
(i) Field winding of the proposed model, such a study has been performed
(ii) c/-axis damper winding
(iii) (/-axis damper winding applying the procedure described in Section 4, and the
results are shown in Figs. 6-10. These Figures show the
armature, field and damper currents, torque, speed and

6 Conclusions
(a) The d-q-0-axes model leads to simplified equations
under balanced conditions of operation. Under unbalanced
operating conditions, the resulting equations require further
transformation. 3 - 4 The solution of such equations is laborious
unless some simplifying assumptions are made.
-4 (b) The <x-j8-0 axes model, which has been found to be
008 016 0-24 0-32 more convenient under certain unbalanced conditions of
time, s operation, results in differential equations with variable
Fig. 8 coefficients. Assuming speed to be constant, an approximate
Variation of torque for single-phase fault with reclosure solution has been suggested.4-5
PROC. IEE, Vol. 118, No. 1, JANUARY 1971 157
(c) If the variation of speed under transient conditions is where R = Ra for an unsymmetrical fault at the machine
taken into account, the equations of performance in models terminals. The rotor voltages will be
using transformed quantities become nonlinear, and a com-
puter solution using numerical methods becomes necessary.
(d) A 3-phase model, which uses direct physical parameters, 0 = P+kd + Xkd'"kd > 02)
is well suited for simulation on a computer, and fault con-
ditions can be simulated more elegantly. J
After substituting expressions for flux linkages in terms of
(e) The parameters of the 3-phase model are physical values,
and it is not necessary to go through complex parameter currents and relevant inductances, in matrix notation, the
transformation as in the case of the other models. performance equations can be written as
(/) The analysis can easily be extended to cases where more
e = Rsl + Lxpl (13)
than two additional rotor circuits have to be taken into
account. where e = [ - Emi sin 8 Emi cos 8 e0 ef ekd ekg]' . . (14)
(g) The proposed 3-phase model and the algorithm presented
for its solution have one or more of the following advantages d q 0 / kd kq
compared with various methods described in literature:
d —Ra — Re v{Lt Le) 0 0 0 ~vLmq
(i) a uniform approach to the study of both symmetrical
and unsymmetrical faults
(ii) a more effective use of computer time for comparable q -v(Ld + Le) - / Re 0 vL
md vL
md 0
accuracy
0 0 0 -R 0 0 0
(iii) retention of the nonlinear model.
Rs =
0 0 0 Rf 0 0

kd 0 0 0 0 Rkd 0
7 Acknowledgments
The authors wish to acknowledge the financial support kq 0 / ? kq J
provided by the National Research Council of Canada
grant A-3727 under which this work was carried out. (15)
and
d q 0 J kd kq

## 8 References -Ld - Le 0 0 Lmd Lmd 0

1 ADKINS, B.: 'Transient theory of synchronous generators connected
to power systems', Proc. IEE, 1951, 98, Pt. II, pp. 510-523
2 CHANTRILL, R. L.: 'Discussion contribution, ibid., 1951, 98, Pt. II
0 -Lq-L 'e ° 0 0 Lm
p. 527
3 CONCORDIA, c.: 'Synchronous machines' (Wiley, 1951) 0 0 -L 0 0 0
4 HWANG, H. H.: 'Unbalanced operation of a.c. machines', IEEE
Trans., 1965, PAS-84, pp. 1054-1066 L, =
5 HWANG, H. H.: 'Unbalanced operation of three phase machines ~Lmd 0 0 hf Lmd 0
with damper circuits', ibid., 1969, PAS-88, pp. 1585-1593
6 DUNFIELD, j . c , and BARTON, T. H. : 'Effect of m.m.f. and permeance
harmonics in electrical machines', Proc. IEE, 1967, 114. (10), kd ~Lmd 0 0 Lmd Lkkd 0
pp. 1443-1450
7 PARK, R. H.: 'Definition of an ideal synchronous machine', Gen.
Elect. Rev., 1928, 31, p. 531 kq 0 0 'kkq-i
8 CHING, Y. K., and ADKINS, B.: 'Transient theory of synchronous
generators under unbalanced conditions', Proc. IEE, 1954, 101, (16)
Pt. IV, pp. 166-182
9 MARTIN, D. w.: 'Runge-Kutta methods for integrating differential In eqn. 16, element L133 = — LQ for an unsymmetrical
equations on high speed digital computers', Comput. J., 1958, 1,
pp. 118-123 fault at the machine terminals. Equations of performance as
10 CLARKE, E. : 'Circuit analysis of a.c. power systems—Vol. 1' (Wiley, indicated above are differential equations with constant
1964) coefficients. They will be linear or nonlinear, depending on
11 MEHTA, D. B., and ADKINS, B. : 'Transient torque and load angle of
a synchronous generator following several types of system dis- whether the speed is constant or varying. A closed-form
turbance', Proc. IEE, 1960, 107 A, pp. 61-74 solution cannot be obtained in the later case.
12 HARLEY, R. G., and ADKINS, B. : 'Calculation of the angular back
swing following a short circuit of a loaded alternator', ibid., 1970,
117, (2), pp. 377-386 9.2 a-^-0 model
13 ADKINS, B.: 'The general theory of electrical machines' (Chapman
& Hall, 1964) To study certain fault conditions, it has been found
convenient 3 ' 4 to use a different transformation than the
d-q-axes transformation. Using a set of orthogonal reference
axes a and j8, the a axis being rigidly fixed to phase a, and
9 Appendixes the /8 axis 90° ahead in the direction of rotation, we have 10
9.1 d-q-0 model
Using Park's transformation, the voltage equations of
the generator connected to an infinite busbar through a
short transmission line can be written as follows. Under
balanced conditions, the machine voltages, in terms of 0o = K0 a + 06 + 0c)
infinite busbar voltages, are
The relationships between d-q and a-j8 variables are (see
ed = — Emi sin S idRe + Le(pid) - vLJq Fig. 1):
(10)
eq = + Emi cos S iqRe vLeid + Le(piq) fa = 0« cos e + 0e s i n 0
ipq = — 0 a sin 6 + i/»p cos 6
Machine voltages can be expressed in terms of flux linkages
as Similar relationships hold good for currents and voltages.
Using the above relationships, the equations of performance
e
d = P^d — VllJq — RJd can be derived from the d-q-axes model already discussed.
Under balanced conditions, machine voltages in terms of
Under unbalanced conditions, (11) infinite busbar voltages are:
e e e e
o — \$( a + b + c) *a = - Emi s i n "t + '«Re + Le(pi
<?p = Emi cos wt + / p /? e + Le(pi\$)
158 PROC. IEE, Vol. 118, No. 1, JANUARY 1971
In terms of flux linkages, the machine voltages are: From eqn. 7,
/4> = e
^3 =
P'Ap ~~ i\$Ra
J ) (25)

## Under unbalanced conditions, (20)

where
V = [—Emi sin cot -Emi sin (cot — 2TT/3)

= PlfjQ — l'QR
-£"„„• sin (tot + 2TT/3) e 7 ekd ekq]'
and
where R = Ra for an unsymmetrical fault at the machine
terminals. The rotor voltages are the same as those given i?m = diag [Ra + i?e, /?„ + Re, Ra + i?e, -Rf, -Rkd, -Rkq]
by eqn. 12. Theflux-linkageequation, in matrix notation, is All the elements of A matrix are zero except Aitii A22, ^3,3*
which are equal to Le.
From eqns. 24 and 25,
where [i/ra fa ip0 -jif ifjkd if>kq]'
Lpl -ApI=V + RmI - pLI
['a 'p *0 '/ hd kg]' (26)
Lmpl = V + RJ-pLI
and
a p 0 f kd kq

—La — Ma cos 20 —Ma sin 20 0 Lmd cos 0 Lmd cos 0 -Lmq sin 0

—Ma sin 20 -La+Ma cos 20 0 Lmd sin 0 Lmd sin 0 LOT(? cos 0

0 0 -L 0 0 0 (21)
L, =
-Lmd cos 0 -Lmd sin 0 0 Lff Lmd 0

## kq Lmq sin Lmq cos 0 0 ••kkq

where Lfl = (Ld + L(?)/2,Mfl = (Ld - Lq)/2, L2 3>3 = - Lo where Lm is the modified inductance matrix having the same
for an unsymmetrical fault at the machine terminals. elements as the L matrix defined by expr. 4, except for the
The performance equations obtained from this transforma- following:
tion are differential equations with variable coefficients. Hence
the advantage of simplification as obtained in d-q trans- Lm\, 1 = - La0 - Le - La2 cos 20
formation is lost. If the speed varies, the equations also Lm2,2 = - LaQ - Le - La2 cos (20 + 2TT/3)
become nonlinear. Using a modified a-fi-0 transformation,
Hwang4 has suggested an approximate method for solving Lm3i3 = - La0 - Le - La2 cos (20 - 277-/3)
the equations under constant-speed conditions. Eqn. 26 yields
In spite of the above mentioned drawback, it has been
claimed4 that this transformation provides a uniform approach pi = L ~ l V + L ~ l R m I —L ~ x p L I . . . . ( 9 )
to study both symmetrical and unsymmetrical faults.
9.5 Digital simulation
9.3 Torque equations 9.5.1 <x-|3-0 model
The equation of motion is given by eqn. 5. In the The iterative algorithm for this model is exactly similar
J-^-0-axes model,13 to that of the 3-phase model described in Section 3. The
equations to be considered are eqns. 19, 20, 21 and 23. The
<x-j8-axes voltages have to be computed at every step using
eqn. 19, after calculating current derivatives making use of
Substituting the relationships indicated by eqn. 18 and similar an equation similar to that of eqn. 9.
expressions for currents in the above, the expression for torque
in the a-/3-0-axes model can be shown to be 9.5.2 d-q-0 model
The equations to be considered are eqns. 10-16 and
7;=^(0«'3-00/«) (23) eqn. 22.
Eqn. 13 can be rewritten as
Substituting the relationships given by eqn. 17 for flux pi = Lf'e - (27)
linkages and similar relations for currents in eqn. 23, the
torque expression in terms of 3-phase armature quantities which gives the differential equations to be solved. Here, Lx
can be obtained as has to be inverted only once, and Rs is required to be formed
at every step since its elements depend upon speed, which
Te = will be varying during transient conditions.
In the 3-phase, and the a-/?-0 models, the resulting solutions
The above expression can also be obtained from the electro- are time dependent, and hence are periodic, whereas, in the
mechanical energy relationships. d-q-0 axes model, the steady-state armature currents are
aperiodic. Hence, for comparable accuracy, the integration-
Equations for calculating current derivatives in
step length for the former simulations must be between one-
9.4
the 3-phase model third and one-sixth of the step length required in the d-q-0-
axes model.
9.6 System parameters
(24) The parameters in per-unit form of the 30MVA syn-
chronous generator1 under study are as follows:
where pL is the derivative of L matrix. The L matrix is
defined by eqn. 4. = 314rad/s = 1-Op.u.
PROC. IEE, Vol. 118, No. 1, JANUARY 1971 159
9.6.1 Transmission line ^ =i-46g
Le = 0 15p.u. Re = 0-0075p.il. Lkkd = 1-4058
Lkkq -0-6535
9.6.2 Machine L m<? = 0-602
(a)Stator
(//) 3-phase model
L*- 0-707 J?/ = 0-000426
r-nniL /?*,= 0-00406
0
L _o:747s J?M = 0-00526
A»o - 0-7415 M r = 0-8919
M,o=-0-3335 M =0-40 3
Aa = 00026 L = 0-93719
(b) Rotor Lkkkdq = 0-43566
(/) Transformed parameters H =
Rf = 000064 ^"'Int14
J? w = 0-0079 ^IH-"0
J?^ = 0 0061 . _Em, Emi
O r
^ = 1-338 ^ - A V Z ^