1

Classical Model
1.0 Assumptions
We will cover Sections 2.5, 2.5.1 (but not, yet, 2.5.2, 2.6),
and 2.7.
Your text, section 2.5, considers a very simplified
representation of a synchronous machine, given in Fig. 1.

Fig. 1
In this model,
E
is the internal machine voltage and
t
V

is the machine terminal voltage. Both of these quantities
are phasors. If the only load in the network is connected
directly across the machine terminals, then the terminal
voltage phasor
t
V
is the reference, i.e.,
0 Z =
t
t
V V
.


X
d

o Z = E E

t
V
2

This is the so-called classical machine model and is based
on the following assumptions:
1. P
au
=T
au
(so we neglect the error introduced by the
power form of the swing equation, due to the fact that

R
).
2. P
m
, the mechanical power into the machine, is
constant. Therefore we consider that the governor is
blocked, and that the effect on mechanical power for
the non-regulated machine that we saw last time is too
small for the kind of frequency changes of interest to
us.
3. There is no damping.
4. Regarding voltages:
a. Internal machine voltage
E
leads main field MMF
r
F
by 90.
b. Terminal voltage
t
V
leads stator MMF s
F
by 90.
The implication of the above two assumptions is that
E
leads
t
V
by the same angle that
r
F
leads s
F
, which
is the torque angle .
3

5.
E
is constant, therefore we neglect the action of the
excitation system.
6. X
d
is the reactance of the direct axis under transient
conditions.
We will consider the case when the synchronous
generator is connected to an infinite bus. An infinite bus
is a bus in which the voltage magnitude and the voltage
frequency are constant.
Infinite buses do not really exist in a power system, but
buses having machines with very large inertias connected
to them tend to behave similar to an infinite bus under
some kind of disturbances. For example, if we were
studying the stability behavior of a 10 MVA machine, and
there was another 1000 MVA machine close by, it would
not be unreasonable to model the 1000 MVA machine as
an infinite bus.
A one-line diagram of a synchronous generator
connected to an infinite bus through a transmission line
of impedance Z
TL
is provided in Fig. 2. The transient
reactance X
d
is not shown but is implied.
4


Fig. 2
We will assume that the synchronous machine of Fig. 1
has some station load which we can model as a constant
impedance Z
S
. The circuit diagram corresponding to Fig. 2
becomes, then, as in Fig. 3.

Fig. 3
Note that the voltage
V
of Fig. 3 is assumed to be the
reference.


X
d

o Z = E E

Z
S

Z
TL

t
V

0 Z =V
V



E
t
V

Z
TL

5

We can convert the circuit of Fig. 3 into a two-node
network which will be more convenient to analyze. This
can be done (for this simple case but not in general)
using a Wye to Delta (Y-) transformation.
Consider the networks of Fig. 4, where the one on the
left has the same topology as the network of Fig. 3.

Fig. 4
If the Fig. 4 networks are equivalent, then the impedance
seen between any two terminals, with the third terminal
open, must be the same. This fact provides 3 equations:
) //( :
2 1 b a c
Z Z Z Z Z b a + = +

) //( :
3 2 c b a
Z Z Z Z Z c b + = +

) //( :
3 1 c a b
Z Z Z Z Z c a + = +




Z
1
=jX
d

Z
3
=Z
S

Z
2
=Z
TL

c
b
Z
c
=1/y
12

Z
b
=1/y
10

Z
a
=1/y
20
c
a b

a
6

Solving these equations for Z
a
, Z
b
, and Z
c
results in
1
3 1 3 2 2 1
Z
Z Z Z Z Z Z
Z
a
+ +
=

2
3 1 3 2 2 1
Z
Z Z Z Z Z Z
Z
b
+ +
=

3
3 1 3 2 2 1
Z
Z Z Z Z Z Z
Z
c
+ +
=

The values of our network, as indicated in Fig. 4, are
Z
1
=jX
d
, Z
2
=Z
TL
, and Z
3
=Z
S
. Substituting these into the
above equations, and taking the inverse to obtain
admittances, results in
S d S TL TL d
d
d
S d S TL TL d
a
Z jX Z Z Z jX
jX
y
jX
Z jX Z Z Z jX
Z
' '
'
'
' '
20
+ +
=
+ +
=

S d S TL TL d
TL
TL
S d S TL TL d
b
Z jX Z Z Z jX
Z
y
Z
Z jX Z Z Z jX
Z
' '
' '
10
+ +
=
+ +
=

S d S TL TL d
S
S
c
Z jX Z Z Z jX
Z
y
Z
Z Z Z Z Z Z
Z
' '
12
3 1 3 2 2 1
+ +
=
+ +
=

With the admittance expressions, we may construct the
Y-bus using the usual rules for Y-bus construction:
7

1. The matrix is symmetric, i.e., Y
ij
=Y
ji
.
2. A diagonal element Y
ii
is obtained as the sum of
admittances for all branches connected to bus i,
including the shunt branch, i.e., where we emphasize
that y
ik
is non-zero only when there exists a physical
connection between buses i and k.
3. The off-diagonal elements are the negative of the
admittances connecting buses i and j, i.e., Y
ij
=-y
ji
.
These observations enable us to formulate the
admittance matrix very quickly from the network based
on visual inspection. We write down the Y-bus for the
network on the right-hand-side of Fig. 4, repeated here
for convenience in Fig. 5.

(

=
(

+
+
=
22 21
12 11
20 12 12
12 10 12

Y Y
Y Y
y y y
y y y
Y


Z
c
=1/y
12

Z
b
=1/y
10

Z
a
=1/y
20
c
a b

Fig. 5
8

In the previous equation, we introduced uppercase
nomenclature, Y
11
, Y
12
, Y
21
, and Y
22
for Y-bus elements,
which differ from the lower case nomenclature used for
admittances y
10
, y
12
, and y
20
. Both uppercase and
lowercase values are complex.
Lets now insert the circuit obtained from the Y-
transformation, Fig. 5, into the system of Fig. 3, as shown
in Fig. 6. The current injections have also been denoted
in Fig. 6.

Fig. 6
The Y-bus relates the nodal current injections to the
nodal voltages, according to eq. 2.40 in your text,
repeated here:
o Z = E E

0 Z =V
V

Z
b
=1/y
10

Z
a
=1/y
20
Z
c
=1/y
12



a

1
I

2
I

9

(

=
(

V
E
Y Y
Y Y
I
I
22 21
12 11
2
1
(1)
Now recall that in per-unit, the power injected into the
network from the source on the left of Fig. 6 is expressed
as
*
1 1 1 1
I E jQ P S = + =
(2)
However, by (1), we can express the current injected as
V Y E Y I
12 11 1
+ =
(3)
Substitution of (3) into (2) results in
( )
*
12
* *
11
2
* *
12
* *
11
*
12 11 1 1 1


Y V E Y E
V Y E E Y E
V Y E Y E jQ P S
+ =
+ =
+ = + =
(4)
Now let
12 12 12
11 11 11
u Z =
+ =
Y Y
jB G Y
(5)
Substituting (5) into (4) results in
10

( )
12 12
*
11 11
2
1 1 1
u Z + = + = Y V E jB G E jQ P S
(6)
Replacing the voltages in the second term with their
polar representation, we get
( )
( )
12 12 11
2
11
2
12 12 11 11
2
1 1 1

0
u o
u o
Z + =
Z Z Z + =
+ =
Y V E B E j G E
Y V E jB G E
jQ P S
(7)
P
1
is the real part of the expression in (7), and Q
1
is the
imaginary part. The swing equation needs only the real
part, and so concentrating on that, we get:
( )
12 12 11
2
1
cos u o + = Y V E G E P
(8)
Now recall that cos(x)=sin(x+/2). Using this in (8):
( ) 2 / sin
12 12 11
2
1
t u o + + = Y V E G E P
(9)
Define =
12
-/2, and substituting into (9), we obtain:
( ) o + = sin
12 11
2
1
Y V E G E P
(10)
Defining P
C
=|E|
2
G
11
and P
M
=|E||V||Y
12
|, (10) becomes
11

( ) o + = sin
1 M C
P P P
(11)
This function is plotted in your text as Fig. 2.9, and it
appears as in Fig. 7 below.

Fig. 7
Some things to note about Fig. 7:
- The generator must supply both the transferred
power, P
M
, and the dissipated power at the station, P
C
.
- Based on (10), the maximum power generation occurs
at -=/2. But we defined (previous page) =
12
-/2;
substitution yields -
12
+/2=/2 =
12
, as indicated
in Fig. 7.
P
1

P
C
+P
M

P
C


12

12

Example 2.3:
This example is worked in your text but I want to clarify
some parts of it. The system appears in Fig. 2.10 of your
text and Fig. 8 below.

Fig. 8
The problem states that H=5 seconds, X
d
=0.20 pu, the
machine is delivering 0.8 pu power at a terminal voltage
of |V
t
|=1.05 pu, all data is given on the machine power
base, and all resistances are neglected. It is desired to
obtain the equation of motion for the machine rotor. No
disturbance is specified, and so we want to determine
the swing equation for the given topology. Doing so is
useful for studying the machine for a small network
disturbance.


E
0 0 . 1 Z = V

Z
1
=j0.4




Z
2
=j0.4
Z
t
=j0.1
t
V

13

Lets understand better what this problem is asking us to
do. We are trying to express the swing equation below in
a way that it can actually be solved.
eu mu
P P t
H
= ) (
2
Re
o
e

(12)
We know the power delivered is 0.8 pu. Since we assume
no losses in the generator, this means P
mu
=0.8 pu, and
based on the classical model assumptions, we assume
that this mechanical power into the machine remains
constant.
What we do not know at this point, and what we need to
determine, is an expression for the electrical power out
of the machine during a small disturbance condition. And
this expression needs to be written in terms of as the
unknown if we are to have any hope of solving (12). So
our initial goal is to express P
eu
as a function of .
To solve this problem, one must first understand that Fig.
8 omits the transient reactance, X
d
. We show how this is
included in Fig. 9.
14


Fig. 9
We can get an equivalent impedance between the
internal voltage and the infinite bus according to
( )
2 1 12
// ' Z Z Z jX Z
t d
+ + =
(13)
where the notation // means paralleled with.
Substituting the numerical values results in Z
12
=j0.5. This
means that y
12
=-j2.0. We can then construct the Y-bus:
(

=
(

=
(

=
0 . 2 0 . 2
0 . 2 0 . 2
22 21
12 11
22 21
12 11
j j
j j
Y Y
Y Y
Y Y
Y Y
Y
(14)
We found in (10) and (11) that
( ) o + = sin
12 11
2
1
Y V E G E P
(10)
( ) o + = sin
1 M C
P P P
(11)


E
0 0 . 1 Z = V

Z
1
=j0.4




Z
2
=j0.4
Z
t
=j0.1
t
V



X
d
=0.2
15

where P
C
=|E|
2
G
11
and P
M
=|E||V||Y
12
|. Noting that our
system is lossless, G
11
=0 implies P
C
=0, and

12
=/_Y
12
=/_j2.0=/2.

Therefore, =
12
-/2=0. So (11)
becomes:
o sin
1 M
P P =
(15)
where P
M
=|E||V||Y
12
|. We do know that |V|=1.0 and
|Y
12
|=2.0. Therefore
o sin | | 2
1
E P =
(16)
One problem we are faced with at this point is that we
do not know |E|. So what do we do?
This is actually a quite typical situation in stability
analysis, even for large-scale multi-machine models. The
situation is: we know terminal conditions of each
machine at t=0 because a time-domain simulation is
always initiated from a steady-state solution. We usually
call this steady-state solution the power flow.
Remember, a power flow only tells us
- the complex power out of the machine terminals, and
- the voltage and angle at the machine terminals.
16

In order to perform time-domain simulation of the
system, beginning from the initial conditions at the
machine terminals provided by a power flow program,
we will always need to back-calculate from the
terminal conditions in order to determine parameters
necessary to initialize the machine model internal
dynamics (which are represented for time-domain
simulation but not for power flow).
Thus, an important step for time-domain simulation is
always to perform this back calculation. For machine
models we will study later in the course, this effort has
some complexities. For the classical machine model, the
effort is straightforward. Lets do it.
The strategy for performing the back calculation is
fundamentally the same, however, independent of the
machine model used. The power flow tells us the power
flow and voltage magnitude and angle at the terminals.
We use this information to get the current out of the
machine. Then we back calculate using the current to
and machine internal impedances to obtain machine
internal voltages.
17

In this case, however, the terminal conditions we know
are power out and voltage magnitude. But we do not
know the machine terminal voltage angle which we do
need. The reason we do not know this is because we
have not done a power flow calculation. So we need to
do that. In this case, it is simple to obtain by inspecting
the circuit, as given in Fig. 10, which is a circuit diagram
for the one-line of Fig. 9.

Fig. 9

Fig. 10


E
0 0 . 1 Z = V

Z
1
=j0.4




Z
2
=j0.4
Z
t
=j0.1
t
V



X
d
=0.2


E
0 0 . 1 Z = V

Z
1
=j0.4




Z
2
=j0.4
Z
t
=j0.1
t
V



X
d
=0.2
18

We want to find E and we could do so if we knew the
current. We can find the current if we can identify
magnitude and angle for two different voltages. We
know magnitude and angle at the infinite bus, and we
know terminal voltage magnitude, so we need to obtain
terminal voltage angle.
This we can do by recognizing that our equation (15) also
applies to the portion of the circuit between terminal
voltage and infinite bus (there is no reason why it should
not since the circuit is identical, a voltage at either end of
a lossless line).
t M
P P u sin
1
=
(15)
where we notice that we replaced the angle with
t
since
it is the angle across the impedances over which we
compute the power flow. P
M
is given by:
P
M
=|V
t
||V|/(X
t
+X
1
//X
2
)=(1.05)(1)/(0.1+0.2)=3.5 (16)
We know the electrical power out is 0.8 pu, therefore
(15) becomes:
t
u sin 5 . 3 8 . 0 =
(17)
19

From (17), we solve to obtain
2286 . 0 5 . 3 / 8 . 0 sin = =
t
u
(18)
And so
t
=0.2306 radians or 13.2147.
Now we can obtain the current, which will be
Z = =
Z Z
=
+

=
285 . 5 8034 . 0 074 . 0 8 . 0
3 . 0
0 1 2147 . 13 05 . 1
) // (
2 1
j
j X X X j
V V
I
t
t
(19)
Now that we have the current, we can compute the
internal voltage according to
Z = + =
Z + Z = + =
09 . 21 1115 . 1 4 . 0 037 . 1
) 285 . 5 8034 . 0 ( 2 . 0 2147 . 13 05 . 1 '
j
j I jX V E
d
t
(20)
Now we may obtain the equation (16):
o sin | | 2
1
E P =
(16)
as
o o sin 223 . 2 sin 1115 . 1 * 2
1
= = P
(17)


20

Now inserting into (12)
eu mu
P P t
H
= ) (
2
Re
o
e

(12)
and recognizing that P
eu
is the same as P
1
, we have
o o
e
sin 223 . 2 ) (
2
Re
=
mu
P t
H

(18)
With H=5,
Re
=377, and P
mu
=0.8, we have:
o o sin 223 . 2 8 . 0 ) (
377
10
= t

(19)
And pushing the constant term to the right-hand-side,
o o sin 8071 . 83 16 . 30 ) ( = t

(20)
Example 2.4:
Here, we want to obtain the equation of motion as in
Example 2.3, but we want to do so for a faulted
condition, where the fault is a balanced three-phase fault
having impedance to ground of j0.1 pu, occurring on one
of the transmission lines very close to the sending end.

21

The situation is illustrated in Fig. 11.

Fig. 11
This circuit may be re-drawn in a more simple fashion as
in Fig. 12.

Fig. 12
We again use a Y- transformation on this. The two
equivalent networks are given in Fig. 13.
E
0 0 . 1 Z = V





Z
2
=j0.2
Z
t
=j0.1
t
V



X
d
=0.2

j0.1


E
0 0 . 1 Z = V

Z
1
=j0.4




Z
2
=j0.4
Z
t
=j0.1
t
V



X
d
=0.2

j0.1
22


Fig. 13

7273 . 2
11 . 0
3 . 0
1 . 0 * 3 . 0 1 . 0 * 2 . 0 2 . 0 * 3 . 0
3 . 0

1
3 1 3 2 2 1
1
20
j
j
j j j j j j
j
Z Z Z Z Z Z
Z
Z
y
a
=

=
+ +
=
+ +
= =

8182 . 1
11 . 0
2 . 0
11 . 0
1
2
10
j
j Z
Z
y
b
=

= =

9091 . 0
11 . 0
1 . 0
11 . 0
1
3
12
j
j Z
Z
y
c
=

= =

The equivalent circuit is shown in Fig. 14.



Z
1
=j0.3

Z
3
=j0.1
Z
2
=j0.2
c
b
Z
c
=1/y
12

Z
b
=1/y
10

Z
a
=1/y
20
c
a b

a
23


Fig. 14
Assuming the internal voltage magnitude remains
constant at the value we found in the last example (an
assumption inherent to the classical machine), which is
|E|=1.1115, and with y
12
=-j.9091Y
12
=j.9091, we have
that
o sin
1 M
P P =
(21)
where P
M
is given by:
P
M
=|E||V|Y
12
|=(1.1115)(1)(.9091)=1.0105 (22)
So (21) becomes
o sin 0105 . 1
1
= P
(23)

o Z = E E

0 Z =V
V

y
10
=-j1.8182
y
20
=-j2.723
y
12
=-j0.9091


a

1
I

2
I

24

Plugging back into (12)
eu mu
P P t
H
= ) (
2
Re
o
e

(12)
(where again, P
eu
=P
1
), we have
o o sin 0105 . 1 8 . 0 ) (
377
10
= t

(24)
Or
o o sin 0958 . 38 16 . 30 ) ( = t

(25)
Finally, it is of interest to consider what happens when
the fault is cleared. In this case, the j0.1 impedance to
ground is eliminated, but so is the faulted line, as
illustrated in Fig. 15.

Fig. 15


E
0 0 . 1 Z = V

Z
1
=j0.4


Z
t
=j0.1
t
V



X
d
=0.2
25

Now the impedance between the voltages is j0.7, and the
transfer admittance is Y
12
=-j1/0.7=-j1.4286, therefore
P
M
=|E||V|Y
12
|=(1.1115)(1)(1.4286)=1.5879 (26)
The swing equation for the post-fault network is then
o o sin 5879 . 1 8 . 0 ) (
377
10
= t

(27)
or
o o sin 8638 . 59 16 . 30 ) ( = t

(28)
Lets summarize the three swing equations for the three
different time frames:
Pre-fault:
o o sin 8071 . 83 16 . 30 ) ( = t


Fault-on:
o o sin 0958 . 38 16 . 30 ) ( = t


Post-fault:
o o sin 8638 . 59 16 . 30 ) ( = t


What do you notice?
Here, we observe that the coefficient of the sin term
varies as follows:

26

- The largest one, 83.807, corresponds to the pre-fault
condition when the transmission system is strongest;
- The smallest one, 38.095, corresponds to the fault-on
condition when the transmission system is weakest.
- The middle one, 59.863, corresponds to the post-fault
condition. Here the transmission system is stronger
than the fault-on condition because the fault has been
eliminated. But it is weaker than the pre-fault
condition because a transmission line has been
eliminated.
Although this example has been for a very simple system,
the observations of the above three bullets apply for real
systems as well.
One concept to clarify here: what does it mean to have a
strong or weak transmission system? Note that the
coefficient of the sin term in the above equations is
H X
V E
2
Re
e


27

The second fraction,
Re
/2H, is present for all three
conditions (pre-fault, fault-on, and post-fault) and so
does not play a role in distinguishing between strong
and weak transmission.
For the first fraction, |E||V|/X, the product |E||V| is the
same among the three conditions, but the parameters X
is different. Here are the differences:
Pre-fault: X=0.3 (see eq. (16) above)
Fault-on: X=1/0.9091=1.1 (see eq. (22) above)
Post-fault: X=1/1.4286=0.7 (see. eq. (26) above)
So we see that
- a strong transmission system (pre-fault) means the
transfer impedance between source and sink is low
(0.3), and
- a weak transmission system (fault-on) means the
transfer impedance between source and sink is high
(1.1).