chapter 9

Multimachine Systems with Constant

Impedance Loads

9. 1 Introduction
In this chapter we develop the equations for the load constraints in a multimachine
system in the special case where the loads are to be represented by constant impedances.
The objective is to give a mathematical description of the multimachine system with the
load constraints included.
Representing loads by constant impedance is not usually considered accurate. It
has been shown in Section 2.11 that this type of load representation could lead to some
error. A more accurate representation of the loads will be discussed in Part III of this
work. Our main concern here is to apply the load constraints to the equations of the
machines. We choose the constant impedance load case because of its relative sim-
plicity and because with this choice all the nodes other than the generator nodes can be
eliminated by network reduction (See Section 2.10.2).

9.2 Statement of the Problem

In previous chapters, mathematical models describing the dynamic behavior·of the
synchronous machine are discussed in some detail. In Chapter 4 [see (4.103) and
(4.138)] it is shown that each machine is described mathematically by a set of equations
of the form
(9.1)
where x is a vector of state variables, v is a vector of voltages, and T'; is the mechanical
torque. The dimension of the vector x depends on the model used. The order of x
ranges from seventh order for the full model (with three rotor circuits) to second order
for the classical model where only wand () are retained as the state variables.
The vector v is a vector of voltages that includes ud , uq , and UF. If the excitation
system is not represented in detail, uF is assumed known; but if the excitation system is
modeled mathematically, additional state variables, including vF, are added to the
vector x (see Chapter 7) with a reference quantity such as VREF known. In this chapter
we will assume without loss of generality that vF is known.
Consider the set of equations (9.1). In the current model developed in Chapter 4, it
represents a set of seven first-order differential equations/or each machine. The number
of the variables, however, is nine: five currents, wand b, and the voltages vd and
uq • Assuming that there are n synchronous machines in the sy.stem, we have a set of
Tn differential equations with 9n unknowns. Therefore, 2n additional equations are
368
 Multimachine Systems with Constant Impedance loads 369

needed to complete the description of the system. These equations are obtained from
the load constraints.
The objective here is to derive relations between vd; and Vq;, i = 1, 2, ... , n, and the
state variables. This will be obtained in the form of a relation between these voltages,
the machine currents i q; and i d; , and the angles 0;. i = 1,2, ... , n. In the case of the flux
linkage model the currents are linear combinations of the flux linkages, as given in
(4.124). For convenience we will use a complex notation defined as follows.
For machine i we define the phasors V; and ~ as
(9.2)
where
Vq; ~ vq;/vr Vd; ~ vd;/vr
t; ~ iq;/vr ld; ~ id;/vr (9.3)
and where the axis q; is taken as the phasor reference in each case. Then we define the
complex vectors V and I by
Vq 1 + j Vd 1 VI
V ~ Vq 2 + j Vd 2 il2

Vqn + j Vd n ~
lq, + j/dl t,
T~ Iq2 + j/d2 t;

t.; + j1dn T" (9.4)

Note carefully that the voltage ~ and the current ~ are referred to the q and d axes
of machine i. In other words the different voltages and currents are expressed in terms
of different reference frames. The desired relation is that which relates the vectors V
and T. When obtained, it will represent a set of n complex algebraic equations, or 2n
real equations. These are the additional equations needed to complete the mathematical
description of the system.

9.3 Matrix Representation of a Passive Network

Consider the multimachine system shown in Figure 9.1. The network has n ma-
chines and r loads. It is similar to the system shown in Figure 2.17 except that the ma-
chines are not represented by the classical model. Thus, the terminal voltages Vi'
i = 1,2, ... , n, are shown in Figure 9.1 instead of the internal EMF's in Figure 2.17.
Since the loads are represented by constant impedances, the network has only n active
sources. Note also that the impedance equivalents of the loads are obtained from the
pretransient conditions in the system.
By network reduction the network shown in Figure 9.1 can be reduced to the n-node
network shown in Figure 9.2 (see Section 2.10.2). For this network the node currents
and voltages expressed in phasor notation are 1;, 1;, ... , T" and ~. V;, ... , v;,
respec-
tively. Again we emphasize that these phasors are expressed in terms of reference
frames that are different for each node.
At steady slate these currents and voltages can be represented by phasors to a com-
370 Chapter 9

-- In
n

Tran smissi on
lL
1
..
Vn • syste m

•
-- I.

V. 0
--- I
Lr

V,

Fig. 9. 1. Mult imachin e system with con stant impedance load s.

mon reference frame . To distingui sh these phasors from tho se defined by (9.2), we will
use the symbols t, and Vi' i = I. 2• . .. , n. to designate the use of a common (net-
work) frame of reference . Similarl y, we can form the matrices i and V. From the net-
work steady-state equations we write
(9 .5)

where

i ~ [n
InJ
V. [:.:]
Vn
(9.6)

and V is the short circuit admittan ce mat rix of the network in Figure 9.2.

9 .3 .1 Network in the transient state

Consider a branch in the reduced network of Figure 9.2. Let this branch, located
between any two nodes in the network. be identified by the subscript k . Let the branch

1 ____
n
n
1.___
••
•
2
1..0-
I
+ +

~_Vn ?_V. 0

?-"
-
F ig. 9.2. Reduced n-port network .
 Multimachine Systems with Constant Impedance loads 371

resistance be 'In its inductance be {k' and its impedance be Zk' The branch voltage
drop and current are v, and i k •
In the transient state the relation between these quantities is given by
k = 1,2, ... , b (9.7)
where b is the number of branches.
Using subscripts abc to denote the phases abc, (9.7) can be written as
k = 1,2, ... , b (9.8)
This branch equation could be written with respect to any of the n q-axis references by
using the appropriate transformation P. Premultiplying (9.8) by the transformation P
as defined by (4.5),
(9.9)
Then from (4.31) and (4.32)

o
(9.10)

Substituting (9.10) in (9.9) and using (4.7).

(9.11 )

which in the case of balanced conditions becomes

(9.12)

It is customary to make the following assumptions: (1) the system angular speed
does not depart appreciably from the rated speed, or w "'" WR and (2) the terms t i are
negligible compared to the terms wti. The first assumption makes the term wtl<il<
approximately equal to XI< iI<' and the second assumption suggests that the terms in II<
are to be neglected.
U nder the above assum ptions (9. t 2) becomes

k = 1,2, ... , b (9.) 3)

Equation (9.13) gives a relation between the voltage drop and the current in one branch
of the network in the transient state. These quantities are expressed in the q-d frame of
reference of any machine. Let the machine associated with this transformation be i.
The rotor angle 0; of this machine is given by

(9.14)
where 0; is the angle between this rotor and a synchronously rotating reference frame.
372 Chapter 9

d.,

ReFerence frome
(mov ing 01 synchronous speed )

Fig . 9 .3. Position of axe s of rotor k with respect to reference frame .

From (9 .13) multiply both sides by I /~; and using (9 .3),

(9 .15)
where the subscript; is added to indicate that the rotor of machine; is used as reference .
Expressing (9.15) in phasor notation,

ilkli} VqkliJ + j Vdklil

(r k Iqklil - x, Im iJ) + jerk Idk1i) + Xk Iqk1iJ) v. + jXdUq k + j/dd
or
k = 1,2, . . . , b, (9 .16)
Equation (9.16) expresses, in complex phasor notation, the relat ion between the
voltage drop in bran ch k and the current in that branch. The reference is the q axis of
some (hypothetical) rotor; located at angle 0; with respect to a synchronously rot ating
system reference, as sho wn in Figure 9.3 .

9 .3.2 Converting to a common reference frame

To obtain general network relationships, it is desirable to express the various

branch quantities to the same reference. Let us assume that we want to convert the
phasor V; = Vq; + j Vdi to the common reference frame (moving at synchronous speed).
Let the same voltage, expressed in the new notation, be V; = VQ; + j VD; as shown in
Figure 9.4.
From Figure 9.4 by inspection we can show that

VQi + j VD ; = (Vq; cos 0; - Vd; sin 0;) + j( Vq; sin 0; + Vdi cos 0i)
or
(9.17)

Now convert the network branch voltage drop equation (9.16) to the system reference
frame by using (9 .17) .
- -j~ .
Vke I zJke-j~;

or
Vk ZJk k = 1,2, . .. , b (9 .18)
where b is the number of branches and Zk is calculated based on rated angular speed.
Comparing (9.18) and (9 .5) under the assumptions stated above, the network in the
transient state can be described by equations similar to those describing its steady-state
 Multimachine Systems with Constant Impedance loads 373

di

V
--- --- - --
Di

-- .i- >:

qi

Fig. 9.4 . Two frames of reference for phasor quantities for a voltage Vi'

behavior. The network (branch) equations are in terms of quantities expressed to the
same frame of reference , conveniently chosen to be moving at synchronous speed (it is
also the system reference frame) .
Equation (9.18) can be expressed in matrix form
(9.19)
where the subscript b is used to indicate a branch matrix. The inverse of the primitive
branch matrix Zb exists and is denoted rb, thus
(9 .20)
Equation (9 .20) is expressed in terms of the primitive admittance matrix of a passive
network . From network theory we learn to construct the node incidence matrix A
which is used to convert (9.20) into a nodal admittance equation
i = (A'YbA)V ~ YV (9.21 )
where Y is the matrix of short circuit driving point and transfer admittances and
I if current in branch p leaves node q
- I if current in branch p enters node q
o if branch p is not connected to node q (9.22)
with p = I, 2, . .. , band q 1,2 , .. . , n .
Since Y - 1 ~ Z exists,
(9.23)
where Z is the matrix of the open circuit driving point and transfer impedances of the
network. (For the derivation of (9 .21)-(9.23), including a discussion of the properties
of the Y and Z matrices, see reference [I]. Chapter II .)

9.4 Converting Machine Coordinates to System Reference

Consider a voltage "ab ri at node i. We can apply Park's transformation to this volt-
age to obtain "dqi' From (9 .2) this voltage can be expressed in phasor notation as V;,
using the rotor of machine i as reference. It can also be expressed to the system refer-
ence as Vi' using the transformation (9 .17).
374 Chapter 9

Equation (9.17) can be generalized to include all the nodes. Let

ej 6 1 0 0

T= 0 ej 62 0 (9.24)

0 0 ejhn

VOl + jVD I Vql + jVd1

V V0 2 + jVD 2 V= Vq2 + jVd2 (9.25)

Then from (9.2), (9.14), (9.17), and (9.25)

V = TV (9.26)
Thus T is a transformation that transforms the d and q quantities of all machines to the
system frame, which is a common frame moving at synchronous speed.
We can easily show that the transformation T is orthogonal, i.e.,
T:" = T* (9.27)
Therefore, from (9.26) and (9.27)
v = T*V (9.28)
Similarly for the node currents we get
i = TI I = r-t (9.29)

9.5 Relation between Machine Currents and Voltages

From (9.22) i = YV." By using (9.29) in (9.22),
rr = YTV (9.30)
Premultiplying (9.30) oy T:"
(9.31)
where
(9.32)
and if M -I exists,
(9.33)
Equation (9.33) is the desired relation needed between the terminal voltages and
currents of the machines. It is given here in an equivalent phasor notation for con-
venience and compactness. It is, however, a set of algebraic equations between 2n real
voltages Vq" Vd1, ... , Vqn, Vdn, and 2n real currents Iq1, Idl , ... , Iqn, Jdn.

Example 9.1
Derive the expression for the matrix M for an n-machine system.
 Multimachine Systems with Constant Impedance loads 375

Solution
The matrix Y of the network is of the form

Y II ej lJ11 Y 12ej tJ 12

Y= Y21 e j 82 1 Y ej (J22
22
Y2n e j (J 2n
(9.34)
Y ej(Jnl Y ej (Jn2 Ynn e j 6nn
nl n2

and from (9.24)

e -jh l

T (9.35)

From (9.34) and (9.35)

YT

and premultiplying by T- ', we get the desired result

j8 11 j(tJ 12- 6 12)
YII e Y12e
j(821 -6 2 1) j 822
Y21 e Y22 e
(9.36)

To simplify (9.36), we note that

Yi/ce j( (Jik - hi/c) _
-
( Gik cos uik
!
+ B ' t) .( Bik cos ui/c
ik sin ui/c + J
t -
G '!)
ik sin "t«

Now define
FG+B(Oi/C) = FG + B = Gik cos Oi/c + Bi/c sin Oi/c
FB-G(Oi/C) F B- G = Bile cos Oi/c - Gilc sin 0ik (9.37)
Then the matrix M is given by
M = H + jS (9.38)
where Hand S are real matrices of dimensions (n x n). Their diagonal and off-
diagonal terms are given by
S;i e, (9.39)

Example 9.2
Derive the relations between the d and q machine voltages and currents for a two-
machine system.
376 Chapter 9

Solution
From (9 .31) and (9 .38)

r = (H + jS) [Vql ~ j VJ] = (H + jsxv, + jVd )

Vqn + j Vdn
= (HV q - SVd ) + j(SVq + HVJ ) (9.40)
For a two-machine system the q axis currents are given by

/ql] [Gil FG+ B (ol2 )l [VqIJ [ 8\1

[ I q2 = FG+ B(021) G22 J Vq2 - FB_ G(021)

and the d axis currents are given by

8 22
I
f"B _G(OI2)] Vql] +
lVq2
l Gil
FG+ B(021)
FG+B(OI 2~ [VdIJ
G22 J Vd2
We note that a relation between the voltages and currents based upon (9 .33) (i.e., giving
Vql, Vq2• Vdl• and Vd2 in terms of t.; I q2, I dl , and I d2) can be easily derived . It would
be analogous to (9.40) except that the admitt ance parameters are replaced with the
parametersof the Z matr ix of the network .

Example 9.3.
Derive the complete system equations for a two-machine system . The machines are
to be represented by the two-axis model (see Section 4 .15 .3). and the loads are to be
represented by constant impedances.

Solution
The transient equivalent circuit of each synchronous machine is given in Figure
4.16. A further approximation, commonly used with this model, is that x;""'" x; ~ x' .
The network is now shown in Figure 9.5. The representation is similar to that or the
classical model except that in Figure 9 .5 the voltages E; and Ei are not constant.
The first step is to reduce the network to the "internal" generator nodes 1 and 2.
Thus the transient generator impedances '1
+ jx; and r, + jxi are included in the net-
work Y (or Z) matrix. The voltages at the nodes are E; = E;I + jE;, and Ei = E;2 + jE;2.
and the currents are 7; = I ql + j/dl and T;. = I q2 + j/J 2' The relation between them is

I;

Fig . 9.5. Network of Example 9.3.

 Multimachine Systems with Constant Impedance Loads 377

given by an equation similar to (9.40). The equations for each machine, under the as-
sumption that x~ ~ x;, are the two axis equations of Section 4.15.3.
, E" - E~i - (X q ; x:) Iq ;
T qO; di -
, EO..,
T dO; qi £FDI - E;i + (Xdi - x;> ldi
Ti;W i Tm ; - (ldiE~i + Iq;£ ;;) - Diwi
iJ; W; - 1 i = 1,2 (9.41 )

Equations (9.40), with Vi replaced with l:, and (9.41) completely describe the
system. Each machine represents a fourth-order system, with state variables £;i'
£~i' Wi' and 0i'
The complete system equations are given by
-[1 - (x ql - x;) B II ] £~I - (x ql - x;) GII £ ; ,

, EO,
- (x q I - x;) F G+ B ( 12) E; 2 ° °
+ (x q 1 -. X;) F B - G( 12) £;2
r q02 d2 -[ 1 - (X q2 - X~) B 22] £~2 - (X q2 - X~) Gn E; 2
- (X q2 - X~) F C +B(021) £;1 + (X q2 - X~) F B- G(02l) £;1
, E', E FDI - [I - (Xdl - x;) BId £;1
TdOI ql

+ (X dl - X;)[G II £~I + F B- G(ol2 ) £;2 + FG+ B(ol2 ) £;2]

£FD2 - [I - (Xd2 - X~) B22] £;2
+ (X d2 - X~)[FB-G(021) £;1 + FG +B (021) £~I + G22£~2]
Tm, - D I WI - [GIl (£~f + £;f) + FB-G(0I2)(£~1 £;2 - £;, £;2)
+ FG+B(0'2)(E~IE~2 + E~IE~2)]
Tm2 - D2 W 2 - [G22(Ed~ + E;~) + FB _ G(021) (£;1 £;2 - £;2 Edt>
+ FG+B(02I)(E~IE~2 + E;I£~2)]
WI - I b2 = W2 - 1 (9.42)

The system given by (9.42) is not an eighth-order system since the equations are not
independent. This system is actually a seventh-order system with state variables
E;" Ed" £;2' Ed2, WI' W2' and 0 12 , The reduction of the order is obtained from the last
two equations

Furthermore, if damping is uniform: i.e., if DI/T;t = D2/ T;2 = D/T; (or if damping is
not present) then the system is further reduced in order by one, and the two torque
equations can be combined in the form

WI2 = Tm ,
Til
- T
m2
T i2
+ 1(£;" £~I' £;2' £~2' on> - ~
T;
WI2

9,6 System Order

In Example 9.3 it was shown that with damping present the order of the system was
reduced by one if the angle of one machine is chosen as reference. It was also pointed
out that if damping is uniform, a further reduction of the system order is achieved.
We now seek to generalize these conclusions. We consider first the classical model with
zero transfer conductances. We can show that the system equations are given by
378 Chapter 9

Tj;W; + D;wi = L
.i=1
E;EjBi;(sin 0;; - sin o~;)
j-;.;

i = 1,2, ... ,n (9.43)

where the superscript s indicates the stable equilibrium angle. Defining the state vec-
tor x, the vector CT, and the function f by
x' = [WI' W2' ••• ,Wn' (01 - O'i), (02 - ( 2), ... , (on - O~)]
h( (J k) = EpEqBpq [sin ( (Jk + O;q) - sin O;q] k = I, 2, ... , m
m = n(n - 1)/2

and (1 = C x where C is a constant matrix. The system (9.43) may then be written in
the form
x=Ax-Bf(u) (9.44)
where A and B are constant matrices.
The order of the system (9.44) is determined by examining the transfer function of
the linear part (with s the Laplace variable)
W(s) = C(sl - A)-I B (9.45)
This has been done in the literature [2, 3]. Expanding (9.45) in partial fractions and
examining the ranks of the coefficients obtained, the minimal order of the system is ob-
tained. It is shown that the minimal order for this system is 2n - 1. For the uniform
damping case, i.e., for constant D;/T;;, the order of the system becomes 2n - 2 (see
also [4]).
The conclusions summarized above for the classical model can be generalized as
follows. If the order of the mathematical model describing the synchronous machine i is
k;, i = 1,2, ... , n, and if damping terms are nonuniform damping, the order of the sys-
tem is (L7= I k, -1). However, if the damping coefficients are uniform or if the damp-
ing terms are not present, a further reduction of the order is obtained by referring all
the speeds to the speed of the reference machine. The system order then becomes
(L?= 1 k; - 2).
The above rule should be kept in mind, especially in situations where eigenvalues
are obtained such as in the linearized models used in Chapter 6. Unless angle differ-
ences are used, the sum of the column of o's will be zero and a zero eigenvalue will be
obtained (see Section 9.12.4).

9.7 Machines Represented By Classical Methods

In the discussion presented above, it is assumed that all the nodes are connected to
controlled sources, with all other nodes eliminated by Kron reduction (see Chapter 2,
Section 2.10.2). The procedure used to obtain (9.31) assumes that all the machines are
represented in detail using Park's transformation. For these machines we seek a rela-
tion, such as (9.3 l ), between the currents I and the voltages V. The former are either
among the state variables if the current model is used, or are derived from the state
variables if the flux linkage model is used (see [5]).
If some machines are represented by the classical model, the magnitudes of their
internal voltages are known. If machine r is represented by the classical model, the
angle (), for this machine is the angle between this internal voltage and the system refer-
ence axis. In phasor notation the voltage of that node, expressed to the system refer-
 Multimachine Systems with Constant Impedance loads 379

ence, is given by
~ = VQ, +jVD, = E,coso, +jE,sino, (9.46)
At any instant if 0, is known, VQ, and VOr are also known.
Since the voltage E, is considered to be along the q axis of the machine represented
by the classical model, we can also express the voltage of this machine in phasor nota-
tion as
ii: = E, + jO r = 1,2, ... , C (9.47)
where c is the number of machines represented by the classical model. Also from
(4.93) on a per phase base

Dividing both sides by three changes the base power to a three-phase base and
divides each voltage and current by W, converting to stator rms equivalent quantities.
Thus we have

and using (9.47),

(9.48)
Note that E, is in per unit to a base of rated voltage to neutral.
Assuming that the speed does not deviate appreciably from the synchronous speed,
then T, '" P, and from the swing equation (4.90) on a three-phase base
5r = wr - 1 (9.49)

A machine r represented by the classical model will have only w, and D, as state
variables. In (9.49) E, is known, while lqr is a variable that should be eliminated. To
do this we should obtain a relation between I q, and the currents of the machines repre-
sented in detail. Similarly the voltages ~i and ~i of the machines represented in
detail should be expressed in terms of the currents lq; and Id; of these machines and the
voltages E, of the machines represented classically. To obtain the above desired rela-
tions, the following procedure is suggested.
Let m be the number of machines represented in detail, and c the number of ma-
chines represented by the classical model; i.e.,
m + cAn

Let the vectors I and V be partitioned as

[~~] ~:]
lqm + j1dm ~m + j~m
T= ---- - - - - - ~ v= -------- ~ (9.50)
I qm+ I + j/dm+ I Em+ 1 +jO

lqn + jldn En + jO
380 Chapter 9

Then from (9.50) and (9.31)

(9.51)

where in (9.51) the complex matrix M is partitioned. Now since Mill exists, (9.51)
can be rearranged with the aid of matrix algebra to obtain

(9.52)

Equation (9.52) is the desired relation between the voltages of the machines represented
in detail along with the currents of the machines represented classically, as functions
of the current variables of the former machines and the known internal voltages of the
latter group. We note that the matrices Mil' M 12 , M 2 1 , and M 22 are functions of
the angle differences as well as the admittance parameters.

Example 9.4
Repeat Example 9.2 assuming that machine 1 is represented in detail by the two-
axis model and machine 2 by the classical model.
Solution
From (9.37) and using Y12 = f 21 and btl = -021'

M = [~~I~j~1~ r_ ~~~J~~~6~2-J (9.53)

Y12e j(912+o 12) I
I
Y22 e J922

and from (9.53) by inspection

M-I = _1_ e- j 8 11
II fit

Y I 2 e j(8 12 -8 11 +6 12) Y I 2 e j ( 6l r b12) Yt2 eH2612-811)

Y II YII
Yn ej8n _ Yf2 e j (28 12 - 6 11) (9.54)
YII
From (9.50) and (9.52)

_1_ e- j8 11 : '_ ~ ej( 8 12 - 8 11- ol2 )

r., Y II I
I
----------+----------------
~ e j ( 61r 6 11 +612) : Yn e j 622 _ Yf2 'e j (2812 - 6 11) (9.55)
Yl1 I r.,
 Multimachine Systems with Constant Impedance loads 381

or

~I

(9.56 )

Note that the variables needed to solve for the swing equations are only ~I' ~"
and I q 2 •

Example 9.5
Repeat Example 9.3, with machine I represented mathematically by the two-axis
model and machine 2 by the classical model.
Solution
Again the nodes retained are the "internal" generator nodes, and the transient im-
pedances of both generators are included in the network Y (or Z) matrix. The equations
needed to describe this system are (9.41) for generator 1, (9.49) for generator 2, and an
additional set of algebraic equations relating the node currents to the node voltages.
Since the two-axis model retains £; and £; as state variables, it is convenient to
use (9.51). For the two-machine system this is the same as (9.40), with E~ replacing
VI and E~ = £2 + jO replacing V2 • The system is now fifth order. The state variables
for this system are £;" £;" WI' W2' and 0'2' The complete system equations are given by
, e:dl
TqO I [B,,(x q , - x~) - 1]£;, - (x q l - x;)[GIIE;, - F G +B(OI2)£ 2]
, i:
TdOI ql £FDI + [BII(X d , - + (Xd l - XD[GIIE~I + FB - G (o,2)E2 ]
x;) - I]E;,
TjlW I Tm, - Dv», - [G'I(E~r + E;f) + FB- G(ol2 ) £; ,E2 + F G+B(OI2)E;I E2]
~i2W2 Tm2 - D 2W 2 - E 2[FG+ B (021) £;1 - F8_G(021)E~1 + G22E2 ]
hl2 WI - W2 (9.57)

9.8 Linearized Model for the Network

From (9.26) V = TV, where T is defined by (9.24) and V and V are defined by
(9.4) and (9.17). Also from (9.31) T = MV, where M is given by (9.32). Linearizing
(9.31),

(9.58)
382 Chapter 9

where Mo is evaluated at the initial angles 0;0' i = 1,2, ... , n, and Vo is the initial value
of the vector V.
Let 0; = 0;0 + 0;4' Then the matrix M becomes

Y"e
j 811 Y12 e j (812 - h 120 - h 12.1) YlneH81n -0 InO- h Inu)

M (9.59)
Yn1ej(lI n, -OnlO-Onl.l) Y j(lJ n2 - 0n20 - 0n2~)
n 2e YnnejlJnn

The general term m., of the matrix M is of the form Yije

j(8ij-6ijO-6;jA),
thus
m.. = y..e j(8 ij-6ijo) e --j6ij 4
'J 'J

Using the relation cos O;jA "J I, sin 0;;4 "J Oij4, we get for the general term
mij "J Yij e j(8ij+6ijO) (I - jO;jA) (9.60)
Therefore the general term in M A is given by

(9.61)
Thus M A has off-diagonal terms only, with all the diagonal terms equal to zero.

o ~O

~O

y j(8,,1- 6,,10) ~
nl e U"IA o ~O

k-'

(9.62)

"
L
k.1

and the linearized equation (9.58) becomes

~ "
IIA Y II e
j 8 11
I"e
j(81,.-&1,,0)
V.4 L,.
k.1
~
kO
Y
lk
ej(8Ik-6IkO) 0
IleA

y Y
/2A 21 e
j(821 - 6210)
2n e
j(82,.-&2,.O)
P;A L
1e.1
~
leO
y e j(8 21e - '21eO)0
21e 2leA

-j

n
Y: y 0
InA "Ie
j(6,. 1- 6n 10)
Y ,." e
j 8""
~A L
1e.1
~
leO nk e
j(8 nle- &nkO)
"leA

(9.63)
 Multimachine Systems with Constant Impedance loads 383

The set of equations (9.63) is that needed to complete the description of the sys-
tem. A similar equation analogous to (9.63) can be derived relating Y.1 to 1.1 and
IJijti. The network elements involved in this case are elements of the open circuit
impedance matrix Z.
We now formulate (9.63) in a more compact form. From (9.24) let T = To + T.1 to
compute

(9.64)

Similarly, we let T- I i N = No + N4 to compute

(9.65)

Note carefully that T- I T l + T~I and that

:/= o rr.r '
:/= (T-I)A = N A. We can
show, however, that (TO)-I = (T-I)o = No- Thus from M = M o + M A we compute
M o + M A = (No + NA)Y(T o + T A). Neglecting second-order terms,

(9.66)

From matrix algebra we get the following relations,

YIn e j 8 1n

Ynl e j l1n l Ynn e j 6nn

~Y
j(6 11- lJlO) '
lie
= ...
Y",e j(8"I-lJ"O)

r l l e j 8 11
r 21 e j(621-6210)

Ynne j8 nn
(9.67)

Also
384 Chapter 9

(9.68)

From (9.66), (9.67), and (9.68)

MA -j[tlAM o - Motlt.\] (9.69)
and the network equation is given by
I~ = MoVA - j[oAMo - MOoA]VO (9.70)
Note that (9.70) is the same as (9.63).
To obtain a relation between VA and lA, we can either manipulate (9.70) to obtain
VA = MoirA - j[OA - Mo'oAMo]Vo (9.71)
or follow a procedure similar to the above. Define
Q : M- ' = T-1y-IT (9.72)
We can then show that
(9.73)

Example 9.6
Derive the relations between VA and t, for a two-machine system.
Solution
From (9.53) we get for M o

(9.74)

(9.75)
 Multimachine Systems with Constant Impedance loads 385

(9.76)

(9.77)

[Iql + J' 1
A
u
] = [ r II e j011 Vq I A + J. Y II e j 8 11 Vd I A +
dIA
Y 12e j( 8 12 - 6
120 > V
+ J. Y Il e j(812 -!S120) Vd26
q 2A
J
Iq2~ + J'1d2~ Y e j(8 12+ o I20 ) V
l2
. Y j(812+ 6120>
qlA + J 12 e VdlA
Y j 8 22 V
+ 22 e
.
q2A + JY22 e
j822
Vd2~

j Y 12 e j(8 12- cS 120) ( Vq 20 + j Vd 20 ) ] '

- ~12A (9.78)
[ -J. Y 12 e j(8 12+o120) (VqlO + J. VdID )

By separating the real and the imaginary terms in equation (9.78), we get four real equa-
tions between I q 16 , Id I A , I q 26 , and I d 26 and ~16' V d 16, V q2A, Vd2~' and 012A. These are
given below:
Iq t 6 = G" Vqt 6 - B" Vdt6 + Y12cos(0I2 - (120) Vq2~ - Y'2 sin ( 0 12 - ( 120) Vd2~
+ Y)2[sin(OI2 - 0)20) V q20 + cos (0 12 - (120) V d20]012A

Id l A B I I Vq l A + Gil Vd l A + Y I2 sin (012 - (120) V q 2A + Y12 cos (012 - 0120) V d 26

+ Y12 [ -cos(OI2 - 0 120) V q 20 + sin (0 12 - 0120) V d20]012A

I q 2A Y 12 cos (8 1l + 0120) V q l A - Y 12 sin (8 12 + OIlO) Vd l A + G22 V q 2A - 8 22 V d2 A
- Y12 [sin(0I2 + 0120} Vq lD + cos (012 + 0120) Vd lO] 0 12 A
Id 2A Y I2 sin (0 12 + 0120) + Y I2 cos (8 12 + 0120) Vd l A + B 22 Vq2~ +
Vq l A G 22 Vd 2A
+ Y 12 [cos (0 12 + 0 120) Vq 10 - sin (Oil + 0 120 ) Vd 10 ] 0 126 (9.79)

Example 9.7
Linearize the two-axis model of the synchronous machine as given by (9.41) and the
classical model as given by (9.48).
Solution
From (9.41) we get
T~oEdA = - Ed A - (x, - x')/ q6
TdOE;A £FDt1 - £;6 + (x, - X')/ d 6
Tj W t1 Tm 6 - DWA - (ldOE d A + Iqo E ; 6 + EdO/d~ + E;olqA)

06 wt1 (9.80)
From (9.48) we get
(9.81 )
 386 Chapter 9

Example 9.8
Linearize the two-machine system of Example 9.5. One machine is represented by the
two-axis model, and the second is represented classically.
Solution
From (9.79), (9.80), and (9.81) and dropping the a subscripts for convenience,
, E·'
TqOI dl [(Xql - x:)B II ~ I]E dl - Gil (Xql - X~)E~1
- [(xql - x~) Yl2 E2 sin (8 12 - 0120)] 012
, E·'
T dOl ql E FD I + [(Xdl - x;)B II - I]E;1 + (Xdl - X;)GII£dl
- [(Xdl - x:) Y 12E2cos (012 - 0120)]012
T ml - DIWI - 1(£dIOBIl + E;IO G II + Iqlo)£;1 + (EdIOG II - £;loB II + IdIO)Edl
+ Y 12 E2[E; 10 sin (812 - 0120) - E dlOCOS (812 - 0120)] 0121
Tm2 - D 2W 2 - E 2HYI2 cos (012 + 0120)] £;1 - [Y 12 sin (012 + 0120)] Edl
- YI2[£;lOsin«(J12 + 0120) + EdIOCOS«(J12 + 0120)]0121
(9.82)
Equation (9.82) is a set of five first-order linear differential equations. It is of the form
x = Ax + Bu, where

£;1 E~, WI W2 ~12

(Xlii - xi)B II - I (Xlii - xi)G II -(Xdl - xI)Y 12E 2cos(8 12 - ~12()
0 0
T~OI Tdol

-(Xql - xi )G II (Xql - xi)B II - I

0 0
TqOI T;OI
-(£~10811 + E;IOGII + l q lo ) -(£dIOGII - £;loB II + I l110 )
A= -DlfT/, 0
TJl Tjl

-E2YI2COS(012 + ~120) £2 Y 12 sin (0\2 + 0120) YI2£2(E~IOsin«(J12 + 0120) + E~IOCOS(lJ'2 + 0120»)

0 -Dd T/ 2
TJ 2 T/2 r»
0 0 -I -I o
(9.84)

From the initial conditions, which determine £dIO, E~IO' E2 , 1;10, Idlo, and 0120 and from
the network Y matrix all the coefficients of the A matrix of (9.84) can be determined.
Stability analysis (such as discussed in Chapter 6) can be conducted.
We note again (as per the discussion in Section 9.6) that the order of the mathe-
matical description of machine I is four, that of machine 2 is two. The system order,
however, is 4 + 2 - I = 5. If the damping terms are not present, the variables WI
and W2 can be combined in one variable WI2.

9.9 Hybrid Formulation

Where a combination of classical and detailed machine representations exists, a
hybrid formulation is convenient. Let m machines be represented in detail, and c
machines represented classically, m + c = n. Then from (9.58),
 Multimachine Systems with Constant Impedance loads 387

~~
VA =
Vm~
~{~~~}
0

From (9.70)

(9.85)

where the subscript m indicates a vector of dimension m.

By comparing (9.85) and (9.63),

(9.86)

where Km(~A) is an (m x I) vector and Kc(6A) is a (c x 1) vector.

From (9.85) and (9.86)

t:~j [~::~ -i- -~:~[~~j -~~~::]

= j
(9.87)

Therefore
I mA = MOil Vm~ - jKm(6~) (9.88)
from which we get
Vm~ = M OI II t., + jM o l \ Km(t5~)
rc~ = M 021 M,o,', t., + jM o21 M OIII Km(cl A) - jKc(cl A ) (9.89)

Example 9.9
Obtain the linearized hybrid formulation for the two-machine system in Example
9.4.
Solution
From Example 9.2

-
Mo = [rr e 11

12e
i 811

j(812 +6120)

(9.90)
388 Chapter 9

Substituting in (9.89)
- 1 - j 8 11 /- . 1 -j 8 11 -v: Y j(812- cSI20) .t
Vl~ = -Y e 14 + J -y e 20 l2e u12~
II II

or

(9.91)

and
YI2 8 cS
1q24 + J0/d24 = -Y e
j(8 12-8 11+&I20>[(1
ql~ + J°1)
dlA + JO(Vq20 + J°Vd20 )Y'l2e j( 12 - I20 > ul2A
.t ]
II

or - JO(Vql0 + J°V)Y
dlO l2 e
j(8 12+ 6120).t
u12A

1q2A + J°1d2A =
YI 2
-Y e
j(812-811+cSI20)(1
ql4 + J· dlA
/)

II

ry22 j(28 12-8 1l)( V . V) Y e j(8 12 + cS I20>( V ° V )] .t

+ J· l 1
-y e q20 + J d20 - 12 qlO + J dlO ul2A
II

(9.92)
Equations (9.91) and (9092) are the desired relations giving i7;A and /21l in terms of IIA
and bI2~. These complex equations represent four real equations:

~11l = -y' cosOlllqlLl + _yl sinOllldl1l + yYl2 [sin(OIl - 012 + 0120)Vq20

II II II

- cos (011 - 012 + 0120) Vd20]012~

I ° I YI2
Vd l A = - -Y sin 8 11l q l A + -y cosOll/dl4 + -y [COS(O'1 - 012 + 0120) ~20
II II 11

+ sin (0 II - 0 12 + 0120) Vd20 ] Ol2A

Yl2 Yl2 .
Iq24 - COS {012 - 011 + 0120)/q 1A - - sm (8 12 - 811 + 0120)ld1A
Yll Yll

+ {- ~t2 [sin (20 12 - ( 22 ) Vq20 + cos (20 12 - 011) Vd 20 1

22 .

+ Y12 [sin (012 + 0120) Vq l O + cos (0 12 + 0120) VdIOl }012L1

YI 2 ° Y12
Id 2A = -y sln(OI2 - 011 + 0120)/q 1A + - cos (012 - 011 + 0120)/d 1A
II Yll

+{ n2
Y
[cos (20
II
12 - 011) Vq 20 - sin(20 12 - 011) VdlOl

- YI2 [COS (0 12 + 0120) VqIO - sin (6 12 + 0120) VdIO l}O

I2A (9.93)

9.10 Network Equations with Flux Linkage Model

The network equation for the flux linkage description is taken from (9.33) and
(9.72).
(9.94)
This is a complex equation of order n, or 2n real equations.
If the flux linkage model is used, I q and I d for the various machines are not state
 Multimachine Systems with Constant Impedance loads 389

variables. Therefore, auxiliary equations are needed to relate these currents to the flux
linkages. These equations are obtained from Section 4.12. For machine i we have

I.
q'
= ..!.-
,{ q
(1 - L MQ)
,{ q
A. _ L
q'
MQ
{q { Q
A Q.
'

1 (1L-Mtd
I di = td D) L
A di - tdtF AFi
MD
-
L
td{D
MD
A Di i = 1,2, ... .n (9.95)

Equations (9.94) and (9.95) are the desired network equations. Together with the
machine equations they complete the description of the system. While the above
procedure appears to be conceptually simple, it is exceedingly complex to implement.
This is illustrated below. To simplify the notation, (9.95) is put in the form
I q; = (Jqi A qi + (J Q; A Qi

I di = (JdiAdi + (JFiA Fi + (JDiAD; i = 1,2, ... .n (9.96)

The complex vector I thus becomes

I q l + j1dl
I = I q2 + jld 2

= [O"qlAql .~.O"QIAQIJ +j [O"diAdl + 0"~1~F1 + O"DIADlJ (9.97)

(Jqn A qn + (J Qn A Qn (Jdn A dn + (JFn A Fn + (J DnADn

Now the matrix Q in (9.94) is of the form

j BII
Z lI e
Q= ••.
[Z j('nl-6",)
"Ie

«; 2 1" sin ~8.,~ - o'")l

= [ 2"1 cos (~;I - o.d X"" J
QR + jQ, (9.98)
Expanding (9.94),
vq + jV d = (QR + jQ/)(Iq + jld )
= (QR1q - Q/ld) + j(Q/lq + QRld) (9.99)
and substituting (9.97) into (9.99),

v q =
(J qnAqn + (JQnAQn

(9.100)
390 Chapter 9

+
(J dnAdn + (J FnAFn + (J OnAOn

(9.101)
Equations (9.100) and" (9.101) are needed to eliminate Vqi and Vdi in the state-space
equations when the flux linkage model, such as given in (4.138), is used.
The above illustrates the complexity of the use of the full-machine flux linkage
model together with the network equations. Much of the labor is reduced when some
of the simplified synchronous machine models of Section 4.15 are used. For exam-
ple, if the constant voltage behind subtransient reactance is used, the voltages E;i
and Ed; become state variables. The network is reduced to the generator internal
nodes. This allows the direct use of a relation similar to (9.31) to complete the mathe-
matical description of the system model. This has been illustrated in some of the
examples used in this chapter.
The linearized equations for the flux linkage model are obtained from (9.97), which
is linear, and (9.73). Following a procedure similar to that used in deriving (9.100) and
(9.101), we expand (9.73) into real and imaginary terms as follows:
·V~ = Vq~ + jVd~
= (QRo + jQ/o)(lq~ + jld~) - j [O~(QRO + jQIO) - (QRO + jQIO)~~)(lqo + jIdo)
= [QRoIq~ - Q/old~ + (OAQ/O - Q/o~A)Iqo + (~~QRO - QRo~A)Ido]
+ j[Q/olq~ + QRoId~ - (O.1QRO - ·QRo~~)lqo + (O.1Q/O - Q/o~A)Ido] (9.102)
The terms in Iq~, IdA' I qo, and Ido are substituted for by the linear combinations of the
flux linkages given by (9.97).

9.11 Total System Equations

From (4.103) for each synch ronous machine and hence for each node in Figure 9.2,
the following relations apply
ik = - Lkl(R k + WkNk)ik - Lkl"k
Wk = (1/3T j k ) ( - A dk ;qk + Aqk;dk - 3Dk W k + 3Tmk )
l, = Wk - 1 k = I, 2, ... .n (9.103)
where i k = [idkiF/ciDkiqkiQk]', "k = [Vdk -VFk 0 Vqk 01' and the matrices Rk , L, and N, are
defined by (4.74). The whole system is of the form
(9.104)
(see [5, 6, 7, 8, and 9]). Assuming that V Fk and Tmk , k = 1,2, ... , n, are known, (9.104)
represents a set of 7n nonlinear differential equations. The vector x includes all the
stator and rotor currents of the machines, and the vector \' includes the stator voltages
plus the rotor voltages (which are assumed to be known). The set (9.31) provides a
constraint between all the stator voltages and currents (in phasor notation) as functions
of the machine angles. These equations are also nonlinear.
 Multimachine Systems with Constant Impedance loads 391

By examining (9.103) and (9.31) we note the following: The differential equations
describing the changes in the machine currents, rotor speeds, and angles are given in
terms of the individual machine parameters only. The voltage-current relationships
(9.3 J) are functions of the angles of all machines. This creates difficulties in the solu-
tion of these equations and is referred to in the literature as "the interface prob-
lem" [10]. The nature of the system equations forces the solution methods to be per-
formed in two different phases (or cycles). One phase deals with the state of the
network, in terms of node voltages and currents, assuming "known" internal machine
quantities. The other phase is the solution of the differential equations of (9.103)
only. The solution alternates between these two phases. This problem is mentioned
here to focus attention on the system and solution complexities. This problem will be
discussed further in Part III of this work.
Finally, if the flux linkage model is used (for the case where saturation is neglected),
the system equations will be (4.138), (9.100), and (9.10 t). Again the "interface prob-
lem" and the computational difficulties are encountered.

Example 9./0
Give the complete system equations for a two-machine system with the machines
represented by the voltage-behind-subtransient-reactance model and the loads repre-
sented by constant impedances.

Solution
The network constraints for this system are given in complex notation in (9.31) or
in real variables in (9.40), and the machine equations are given in Section 4.15.2. The
machine equations are obtained from (4.234) and (4.270). They are

E;~ = KI;E;; + KuA D;

~; = - f;ldi - Iq;X!, + Ed;
~; = - r;lq; + IdiX:' + E;~ (9.105)
and

T;~;E;J; - E~: - (x q; - x;~)/q;

T:!o;A. D; = E;; - AD; + (x~; - X;f.;)ldi
T~o;E;; = E FD; - (I + Kd;)E;i + Xd;!di + Kd;ADi
1jiW; = Tm ; - Iq;E;: - Id;E;J;
b; = W; - 1 i = I, 2 (9.106)
The network constraints are obtained from (9.40).
The system has ten differential equations, six auxiliary machine equations, and four
algebraic equations for the network (or two complex equations). As per the discussion
in Section 9.6, some differential equations can be eliminated by using ~I - ~2 and
WI - W2 as state variables.
Some of the computational labor can be reduced if the subtransient reactances of
the generators are included in the network Y matrix (or Z matrix). The network equa-
tions would then give relations between the currents I q; and I di, i = I, 2, and the volt-
ages E;~ and Ed;, i = 1,2. The auxiliary equations for ~i and Jt;,i can be omitted. Also
in (9.40), E;: and E;J; should replace ~i and ~;.
392 Chapter 9

9.12 Multimachine System Study

The nine-bus system discussed in Section 2.10 is to be examined for dynamic sta-
bility at the initial operating point given in Section 2.10. Linearized machine equations
are to be used. The loads are to be simulated by constant impedances based on the
initial operating conditions.
The system under study comprises three generators and three loads. A one-line
impedance diagram is given in Figure 2.18. The initial operating system condition,
indicating the power flows and bus voltages, is given in Figure 2.19. Data for the
three generators are given in Table 2.1 (some of which are repeated below for con-
venience).
The synchronous machine models to be used are as follows: classical model for
generator 1, and the two-axis model for generators 2 and 3.

9.12.1 Preliminary calculations

Let the generator terminal voltage be V 1Ji, and the q axis be located at angle o. All
angles are measured from reference. The generator current flags the terminal voltage
by the power factor angle e. The following relations, derived in Section 5.5, are used
(r ,-...; 0) to obtain the data in Table 9.1:

I; + j/x = I /- l/J = (P - jQ)/V tan(a - (3) = xqlr/(V - xqlx)

Table 9.1. Three-Machine System Data

Generator I Generator 2 Generator 3
Quantity Unit
(classical) (two-axis) (two-axis)

H(MW. s/IOO MVA) s 23.6400 6.4000 3.0100

Tj = 2HwB pu 17824.1400 4825.4863 2269.4865
Xd - xd pu 0.0852 0.7760 1.1312
xq - xd pu 0.0361 0.7447 1.0765
T~O s 0 0.5350 0.6000
T~O pu 0 201.6900 226.1900
TdO s 8.9600 6.0000 5.8900
TdO pu 3377.8404 2261.9467 2220.4777
E~o pu 0.7882 0.7679
EdO pu -0.6940 -0.6668
I qo pu 0.6780 0.9320 0.6194
I do pu -0.2872 -1.2902 -0.5615
VqO pu 1.0392 0.6336 0.6661
VdO pu -0.0412 -0.8057 -0.7791
s, elec deg 2.2717° 61.0975° 54.1431 °
E' pu 1.0566

9.12.2 Linearized network equations

The network is assumed to include the transient reactances of the generators. The
network is reduced to the generator internal nodes. At these nodes the voltages are
E~, E~, and E;.
From (9.63) with V replaced with E' and for a three-machine system (using
~'2 = - ~2" ~13 = - ~31)'
 Multimachine Systems with Constant Impedance loads 393

YI 2eJ(' I2 - ' 120) Yllcj('1J -6no) - jEio Y12ej (' 12-'120) - jEio YJ)e H' IJ- 6 IJo)
Yn ej' 22 Y23eJ(' 23 - '230) j £;0 Y2Ie j (' 21-'210) 0
Y32eH'J2 - 6320) Yn ej' )3 0 jE:o Y31e j( 831- '310)

(9.107)
With generator I represented classically, £;A = 0 and E~ £,; and with node I
as the arbitrary reference node = £1 + jO = E, (a constant). E:
Substituting in
(9.107) and using 023 = 013 - 012'

~A Yl2 e j (812 - 6 120' Yl3 e j( 813 - 6130) - jE20 Y12e j( 8 12- 6 120) - jEio YlJ e j (8 13 - 6 130)
E~A
7;/1 Y22e j822 Y23e j( 8 n - 6230) j[E I Y2Ie j( 8 12+6 120) - j Eio Y23e j( 823 - 6230)
EiA
+ £;0 Y23e j( 823 - 6230)]
°12A
1;A
Yn e j( 823 +6230) YBej833 - jE~o Y23e j( 823 +6 230) j[EI Ynejl813 +6130)
+ E~o y 23eH823 + 6230)] 013A

(9.108)
Separating real and imaginary parts and dropping the subscript Ll for convenience,
£;2 £:12 £;J Ed) ~'2 ~I)

If' Y12COS(012 - 0120) - Y12 sin (0 12 - h120) Y" cos (Ou - ~uo) - Y.) sin (01)- ~I)o) Y12[£ ,i20 cos (0 12 - ~120) YuIE,iJocos(Ou - ~IJO) £;2
+ E;20sin (012 - ~120)1 + £~JOsin(Ou - 01)0»
i.; Y12sin «(J12 - 0120) Y1 2cos(012- 0'20) Yu sin (0" - tS DO) Yucos(O'J - ouo) YIZ!£d20 sin (0 12 - 0.20) YIJI£';JO sin (0" - tS uo) £:12
- £;20COS(0I2 - ~l2o») - £;JOcos(Of) - 0,,0»)

1'2 G22 -8 22 Y2J cos(On - 02)0) - Y2)sin(On - 0no) -e, YI2 sin (812 + 0 20)1 Y2JI£ dJO cos (82J - °no) £;]
- £';10 YncOS(02J - 02)0) + £;losin (Ou - 0130)]
- £;JOY2J sin (02)- 0230)
I d2 8 12 G22 YHsin(OH - 02JO) YZJcos(02J - 6B o ) E, Y'2 COS(OIZ + 0 120) Yn[EdJOsin(82J - 0210) EdJ
- E,;JO Yn sin (02J - OlJO) - £;30cos (021 - 0130)]
+ £;10 Y2Jcos(8 2J - 02JO)
If J Yncos(02J + 0])0) - Y2J sin (82J + 0BO) GJ) -B 13 Y2J[Ed20COS(OZJ + 02JO) -£1 YlJsin(OIJ + 0110) hl1
+ £;20sin(OB + °2Jo)1 - Y2JE';20COS(On + 02JO)
- Y2J£;20sin(fl2J + 02J0)
I dJ Y21sin(OB + 0210) Y2Jcos(01J + 0210) BJ) GJ] Yn[Ed 20sin(01J + ono) £1 YUcos(fJll + OUO) Ou
- £;zocos(8 H + (hJo)J - £:110 YH sin (82) + ovo)
+ £;20 Y2.lCOS(OlJ + 02JO)

(9.109)
Equation (9.109) is the desired linearized network equation. It relates the incremental
currents to the incremental state variables E;2' £:12' £;3' £:13' 012' and 013.
9.12.3 Generator equations
From Example 9.7 we obtain the following generator equations (again the subscript
Ll is omitted):
Generator I (classical)

(9.110)
394 Chapter 9

Generators 2 and 3 (two-axis model)

r;o;Ed; = -Ed; - (x q; - x;)Iq;
Tdo;E;; = EFDi - E;i + (Xd; - xI )I d;
Tj;W; = Tm; - D,», - IdioEdi - IqioE;i - E~iOld; - E~ioIqi
&i = Wi i = 2,3 (9.111 )
Again we recall that, to obtain an independent set, the last equations in (9.110) and
(9.111) are combined to give
s; = Wt - Wi ; = 2,3 (9.112)
By using (9.109), t.; t.; Iq2' Id2, I qJ, and IdJ are eliminated from (9.110) and
(9.111). The resulting system comprises nine linear first-order differential equations.
The state variables are E;2, E:n , E;3, EdJ~ WI' W2, WJ, 012, and 013'

9.12.4 Development of the A matrix

The Y matrix of the network, reduced to the internal generator nodes and including
the generator transient reactance, is given in Table 2.6 as the prefault Y matrix. It is
repeated here in Table 9.2. Data for the terms in (9.109) are calculated and given in
Table 9.3.

Table 9.2. Reduced V Matrix for a Three-Machine System

Node 2 3

0.8455 - j2.9883 0.2871 + jJ.5129 0.2096 + j 1.2256

= 1.5399 /79.25° = 1.2434 /80.30°
2 0.2871 + jl.5129 0.4200 - j2.7238 0.2133 + j 1.0879
= 1.5399 /79.25° = 1.1086 /78.91°
3 0.2096 + j 1.2256 0.2133 + j 1.0879 0.2770 - j2.3681
= 1.2434 /80.30° = 1.1086/78.91 0

The coefficients of (9.109) and (9.111) are then calculated. The main system equa-
tions are given below. The incremental currents I q; and I di are calculated from
(9.1.09).

t.; -1.1458 -1.0288 -0.8347 -0.9216 1.6062 1.2642 £;2

ldt 1.0288 -1.1458 0.9216 -0.8347 0.1891 0.0265 E:n
lq2 0.4200 2.7239 0.3434 -1.0541 -1.1484 0.5805 £;3
I d2 -2.7239 0.4200 1.0541 0.3434 2.4914 -0.9666 E d3

i., 0.0800 -1.1058 0.2770 2.3681 0.8160 -1.4414 012

I d3 1.1058 0.0800 -2.3681 0.2770 -0.8305 1.9859 013
(9.113)
The generator differential equations are:
Generator 1 (classical)
WI = 5.6104 x 10-sTmt - 5.6104 x 10-sDIWt - 5.9279 - lO-5lqt
~I = WI
 Multimachine Systems with Constant Impedance loads 395

Table 9.3. Preliminary Calculations for Three-Machine System

Nodes 1-·-2 1-3 2-3
Yij 1.5399 1.2434 1.1086
Oij 79.2544 80.2952 78.9084
oijO - 58.8259 -51.8714 6.9545
Oij - oijO 138.0802 132.1666 71.9540
Yij cos (Oij - oijo) -1.1458 -0.8347 0.3434
Yijsin(Oij - oijo) 1.0288 0.9216 1.0541
Oij + oijO 20.4285 28.4238 85.8629
Yij cos (Oij + oijo) 1.4431 1.0935 0.0800
Yij sin (Oij + oijo) 0.5375 0.5919 1.1058

Generator 2 (two-axis)
E:J2 = -4.9581 x 10-3Ed2 - 3.6923 10-3/q2
X

£;2 = 4.4210 x 10-4EFD2 - 4.4210 x 10- 4£ ; 2 + 3.4307 x 10-4/d2

W2 2.0723 X 10- 4T - m2 2.0723 X 10- 4D2W2 - 1.9314 X 10- 4£;2
+ 2.6736 x 10-4£d2 + 1.4383 X 10-4/ d2 - 1.6334 X 10-4/q 2
~2 W2

Generator 3 (two-axis)
£:13 -4.4210 x 10-3Ed3 - 4.7592 X 10-J/q3
£;3 4.5035 x 10-4E FD3 - 4.5035 x 10- 4£ ; 3 + 5.0944 x 10-4/ dJ
WJ = 4.4063 X 10- 4Tm J - 4.4063 X 10- 4DJW3 + 2.4741 X 10- 4E dJ
- 2.7292 x 10- 4£ ;3 + 2.9380 x 10-4/ dJ - 3.3836 X 10-4/ q3
53 W3 (9.114)
By using (9.113), the currents are then eliminated in (9.114). Combining terms and
using the relation oij = 0; - OJ' we obtain the linearized differential equations for
the three-machine system. The results are shown in (9.115), which is of the form
WI £;2 £;2 W2 £;J £;J WJ 6 11 61)

WI -0.56IOD, 0.6793 0.6099 0 0.4948 0.5463 0 -0.9520 -0.7494 WI

£;2 0 -13.7658 1.4409 0 3.6163 1.1781 0 8.5472 -3.3161 £;2
£:12 0 -15.5076 -150.1554 0 - 12.6793 38.9205 0 42.4023 -21.4333 £;2
W2 0 - 6.5352 -1.1714 -2.0723b 2 0.9552 2.2156 0 5.4592 -2.3385 W2

i;) ,.. 10- 4 0 5.6334 0.4076 0 -16.5675 1.4111 0 -4.2309 10.1170 £;3
i;n 0 - 3.807 J 52.6270 0 -13.1829 -156.9117 0 -38.8349 68.5987 £;3
w) 0 2.9781 3.9766 0 - 10.6238 -4.7247 -4.4063D) -5.2010 10.7116 WJ
~12 10000 0 0 -10000 0 0 0 0 0 612
~I) 10000 0 0 0 0 0 -tOOOO 0 0 ~I)J
0.5610T""
4.421O£'D2
0
2.0723T"'2
+ 10-4 4.5035£'D3 (9.115)
0
4.4063 T"'3
0
0
396 Chapter 9

x = Ax + Bu
where x' = [WI E~2 E:J2 W2 E~3Ed3 W3 Oil 013]
u': = [Tm , E FD2 Tm2 E FD 3 Tm3 ]
The eigenvalues of the A matrix are obtained for the case of D I D2
1.0 pu, using a library computer program. They are
x, -0.002664 + jO.034648 A6 -0.010373
A2 -0.002664 - jO.034648 A7 -0.000455
A3 -0.000622 + jO.022984 As -0.000199 + jO.000129
A4 -0.000622 - jO.022984 A9 -0.000199 - jO.000129
As = -0.016644
All the eigenvalues have negative real parts, and the system is stable for the op-
erating point under study. The dominant frequencies are about 2.1 Hz and 1.4 Hz
respectively. These frequencies are the rotor electromechanical oscillations and should
be very similar to the frequencies obtained in Example 3.4. Thus if we plot Pl2 from
the data of Figure 3.8, we find that the dominant frequency is about 1.4 Hz, which
checks with the data obtained here.
A similar run was obtained for the same data except for D, = D2 = D3 = O.
The eigenvalues are
AI -0.000458 -0.000529 + jO.022983
A2 -0.000281 -0.000529 - jO.022983
A3 -0.010366 -0.002459 + jO.034636
A4 -0.016659 -0.002459 - jO.034636
As = 0
Since this is a special case of uniform damping (D/Tj = 0), the system order is re-
duced by one. The frequencies corresponding to the electromechanical oscillations are
almost unchanged, while the long period frequency has disappeared.

Problems
9.1 If the Y matrix of the network, reduced to the generator nodes, is such that Oij =
90°, i ~ j, derive the general form of the matrix M.
9.2 For the conditions of Problem 9.1, obtain the real matrices for Iq and Id in terms of
Vq and Vd. Compare with (9.40) for a two-machine system with Gil = G21 = O.
9.3 Repeat Example 9.3, using the synchronous machine model called the one-axis model (see
Section 4.15.4).
9.4 Repeat Example 9.5, neglecting the amortisseur effects for the synchronous machine repre-
sented in detail (Section 4.15.1).
9.5 Linearize the voltage-behind-subtransient-reactance model of the synchronous machine.
9.6 Repeat Example 9.8, using the results of Problem 9.5.
9.7 Develop (9.89) for a three-machine system with zero transfer conductances.
9.8 For the nine-bus system of Section 2.10 the dynamic stability of the postfault system (with
line 5-7 open) is to be examined. The generator powers are the same as those of prefault
conditions.
a. From a load-flow study obtain the system flows, voltages, and angles.
b. Ca~culate the initial position of the q axes; lqo, ldo, VqO, VdO, E~o, and E dO for each ma-
chine; and the angles 0120 and 0130.
c. Obtain the A matrix and examine the system eigenvalues for stability.
 Multimachine Systems with Constant Impedance loads 397

References
I. Anderson. Paul M. Analysis of Faulted Power Systems. Iowa State University Press, Ames, 1973.
2. Pai, M. A., and Murthy, P. G. New Liapunov functions for power systems based on minimal realiza-
tions. Int. J. Control 19:401-15, 1974.
3. Willems, J. L. A partial stability approach to the problem of transient power system stability. Int. J.
Control 19:1-14, 1974.
4. Pal, M. K. State-space representation of multimachine power systems. IEEE Paper C 74 396-8. pre-
sented at the Summer Power Meeting, Anaheim, Calif, 1974.
5. Prabhashankar, K., and Janischewskyj, W. Digital simulation of multimachine power systems for sta-
bility studies. IEEE Trans. PAS-87:73-80, 1968.
6. Undrill, J. M. Dynamic stability calculations for an arbitrary number of interconnected synchronous
machines. IEEE Trans. PAS-87:835-44, 1968.
7. Janischewskyj, W., Prabhashankar, K., and Dandeno, P. Simulation of the nonlinear dynamic re-
sponse of interconnected synchronous machines (in two parts). IEEE Trans. PAS-91:2064-77,1972.
8. Van Ness, J. E., and Goddard, W. F. Formation of the coefficient matrix of a large dynamic system.
IEEE Trans. PAS-87:80--84, 1968.
9. Laughton, M. A. Matrix analysis of dynamic stability in synchronous multi-machine systems. Proc.
lEE (British) 113:325-36, 1966.
10. Tinney, W. F. Evaluation of concepts for studying transient stability. IEEE Power Engineering So-
ciety Tutorial, Spec. Publ. 70 M62-PWR, pp. 53-60, 1970.