Vous êtes sur la page 1sur 156

1

Review of Chapter 2, Plus Matlab Examples




2.2 Power in single-phase circuits

Let ( ) ( ) and v t i t be defined as:

( ) ( ) ( ) ( ) cos and cos
m v m i
v t V t i t I t = + = +

then the instantaneous power is give by:


( ) ( ) ( ) ( ) ( )
( ) ( )
( ) ( ) ( )
{ }
( ) ( ) ( ) ( ) ( ) { }
cos cos
cos 2 cos
2
cos 2 cos
cos2 cos sin2 sin cos
m m v i
m m
v i v i
rms rms v v i v i
v v i v v i v i
p t v t i t V I t t
V I
t
V I t
V I t t




= = + +
= + + +

= + +

= + + + +


( ) ( ) ( ) cos 1 cos 2 sin sin 2
v v
p t V I t V I t = + + + +



where
v i
= . The angle is called the power factor angle and is the angle between
the voltage and current at a particular 2-port. Note: the power factor is given by cos .
Note that the cosine is an even function thus the sign of is lost! For an inductive
circuit (memory helper: ELI) voltage leads the current, the carrent lags the voltage and
we call this a "lagging" power factor because the current lags the voltage. For a
capacitive circuit (memory helper: ICE) the current leads the voltage, and we call this a
"leading" power factor. Thus, we always say clearly if the power factor is lagging
(inductive) or leading (capacitive). The lag or lead is the angle by which the current lags
or leads the voltage.
It is also noted that is the impedance angle, thus: ( )
v i
V
V
Z
I I
= = . The
instantaneous power can be expressed in two parts:

( ) ( ) ( )
R X
p t p t p t = +

thus ( ) ( ) cos 1 cos 2
R v
p t V I t = + +

and ( ) ( ) sin sin 2
X v
p t V I t = + .

It is common to define P and Q as follows:


2
( ) ( ) ( ) ( )
cos
R
P avg p t avg p t V I = = =

sin Q V I =

Thus in terms of P and Q, the instantaneous power can be expressed as:

( ) ( ) ( ) 1 cos 2 sin 2
v v
p t P t Q t = + + + +



Thus P is the "average" value of the cosine terms (counting DC as a zero frequency
cosine term) while Q is the "magnitude" of the sine term. It is noted that besides the DC
term, power has twice line frequency components.

Example 2.1

A supply voltage ( ) 100cos v t t = is applied across a load whose impedance is given by
1.25 60 Z =
!
. Determine ( ) i t and ( ) p t . Use Matlab to plot ( ) ( ) ( ) ( ) , , ,
R
i t v t p t p t
and ( )
X
p t over an interval from 0 to 2 .

The Matlab program is shown below:
Vm = 100; thetav = 0; % Voltage amplitude and phase angle
Z = 1.25; gama = 60; % Impedance magnitude and phase angle
thetai = thetav - gama; % Current phase angle in degree
theta = (thetav - thetai)*pi/180; % Degree to radian
Im = Vm/Z; % Current amplitude
wt=0:.05:2*pi; % wt from 0 to 2*pi
v=Vm*cos(wt); % Instantaneous voltage
i=Im*cos(wt + thetai*pi/180); % Instantaneous current
p=v.*i; % Instantaneous power
V=Vm/sqrt(2); I=Im/sqrt(2); % rms voltage and current
P = V*I*cos(theta); % Average power
Q = V*I*sin(theta); % Reactive power
S = P + j*Q % Complex power
pr = P*(1 + cos(2*(wt + thetav))); % Eq. (2.6)
px = Q*sin(2*(wt + thetav)); % Eq. (2.8)
PP=P*ones(1, length(wt)); % Average power with length w for plot
xline = zeros(1, length(wt)); % generates a zero vector
wt=180/pi*wt; % converting radian to degree
subplot(2,2,1), plot(wt, v, wt, i,wt, xline), grid
title(['v(t)=Vm coswt, i(t)=Im cos(wt +', num2str(thetai), ')'])
xlabel('wt, degree')
subplot(2,2,2), plot(wt, p, wt, xline), grid
title('p(t)=v(t) i(t)'), xlabel('wt, degree')
subplot(2,2,3), plot(wt, pr, wt, PP, wt,xline), grid
title('pr(t) Eq. 2.6'), xlabel('wt, degree')
subplot(2,2,4), plot(wt, px, wt, xline), grid
title('px(t) Eq. 2.8'), xlabel('wt, degree')
subplot(111)


3
S =
2.0000e+003 +3.4641e+003i
0 100 200 300 400
-100
-50
0
50
100
v(t)=Vm coswt, i(t)=Im cos(wt +-60)
wt, degree
0 100 200 300 400
-2000
0
2000
4000
6000
p(t)=v(t) i(t)
wt, degree
0 100 200 300 400
0
1000
2000
3000
4000
pr(t) Eq. 2.6
wt, degree
0 100 200 300 400
-4000
-2000
0
2000
4000
px(t) Eq. 2.8
wt, degree



2.3 Complex power

It is observed that the term
*
VI gives the result: (N.B.
v i
= )


*
cos sin VI V I j V I = +

which is identical to
*
S VI P jQ = = + where S is the complex power.

OTHER FORMS for S :
2
2 2 2
*
V
S R I jX I Z I
Z
= + = = .

4
V
I

S
P
Q
Phasor diagrams (V, I) and the power triangle for
an inductive load
V
I

S
P
Q
Phasor diagrams (V, I) and the power triangle for
a capacitive load, leading power factor angle
For example, for an inductive load, the current lags the voltage and the phasor diagrams
would be as shown below:


And for a capacitive load, the current would lead the voltage , and the phasor diagrams
would be as shown below:




2.4 Complex power balance

The total value of S for a circuit is the SUM of S for the components of S .

Example 2.2:

Three impedances in parallel are suppleid by a source 1200 0 V =
!
V where the
impedances are given by:
1
60 0 Z j = + ,
2
6 12 Z j = + and
3
30 30 Z j = . Find the power
absorbed by each load and the total complex power. As a check, compute the power
supplied by the source.

5

Matlab program follows:

V = 1200; Z1= 60; Z2 = 6 + j*12; Z3 = 30 - j*30;
I1 = V/Z1; I2 = V/Z2; I3 = V/Z3;
S1= V*conj(I1), S2= V*conj(I2), S3= V*conj(I3)
S = S1 + S2 + S3
Sv = V*conj(I1+I2+I3)

S1 =
24000
S2 =
4.8000e+004 +9.6000e+004i
S3 =
2.4000e+004 -2.4000e+004i
S =
9.6000e+004 +7.2000e+004i
Sv =
9.6000e+004 +7.2000e+004i

It is observed that the complex power is conserved.

6
P
Q
S
d

Q
C
Example 2.3
2.5 Power factor correction

Adding a capacitor (usually in parallel) with an inductive load can improve the power
factor.

Example 2.3

Two loads
1
100 0 Z j = + and
2
10 20 Z j = + are connected across a 200-V 60 Hz source.
(a) Find the total real and reactive power, the power factor at the source, and the total
current.
(b) Find the capacitance of the capacitor connected across the loads to improve the
overall power factor to 0.8 lagging.



The Matlab program follows:

V = 200; Z1= 100; Z2 = 10 + j*20;
I1 = V/Z1; I2 = V/Z2;
S1= V*conj(I1), S2= V*conj(I2)
I = I1 + I2
S = S1 + S2, P = real(S); Q = imag(S);
PF = cos(angle(S))
thd = acos(0.8), Qd = P*tan(thd)
Sc = -j*(Q - Qd)
Zc = V^2/conj(Sc), C = 1/(2*pi*60*abs(Zc))
Sd = P + j*Qd;
Id=conj(Sd)/conj(V)

S1 =
400
S2 =
8.0000e+002 +1.6000e+003i
I =
6.0000 - 8.0000i
S =
1.2000e+003 +1.6000e+003i
PF =

7
0.6000
thd =
0.6435
Qd =
900
Sc =
0 -7.0000e+002i
Zc =
0 -57.1429i
C =
4.6420e-005
Id =
6.0000 - 4.5000i



Example 2.4

Three loads are connected in parallel across a 1400-V, 60 Hz single-phase supply. Given
that load 1 is inductive, 125 kVA at 0.28 power factor, load 2 is capacitive, 10 kW and
40 kvar, and load 3 is resistive at 15 kW.

(a) Find the total kW, kVA and the supply power factor.
(b) A capacitor is connected in parallel with the loads to improve the power factor to 0.8
lagging. Find the kvar rating of the capacitor and its capacitance in F.

The Matlab program follows:

V = 1400;
S1= 35000 + j*120000; S2 = 10000 - j*40000; S3 = 15000;
S = S1 + S2 + S3, P = real(S); Q = imag(S);
PF = cos(angle(S))
I = conj(S)/conj(V)
thd = acos(0.8), Qd = P*tan(thd)
Sc = -j*(Q - Qd)
Zc = V^2/conj(Sc), C = 1E6/(2*pi*60*abs(Zc))
Sd = P + j*Qd;
Id=conj(Sd)/conj(V)

S =
6.0000e+004 +8.0000e+004i
PF =
0.6000
I =
42.8571 -57.1429i
thd =
0.6435
Qd =
45000
Sc =
0 -3.5000e+004i
Zc =
0 -56.0000i

8
C =
47.3675
Id =
42.8571 -32.1429i


2.6 Complex Power Flow

This is a very important part of chapter 2. Note the diagram below. It shows four
power flows. They are:

1.
12
S is the complex power at node number 1 flowing in (i.e. from node 1 in the
direction of node 2).
2.
12
S is the complex power at node number 1 flowing out (i.e. from node 1 in the
direction away from node 2).
3.
21
S is the complex power at node number 2 flowing in (i.e. from node 2 in the
direction of node 1).
4.
21
S is the complex power at node number 2 flowing out (i.e. from node 2 in the
direction away from node 1).




Thus the power consumed by the impedance Z can be expressed as
12 21 Z
S S S = + .

Important Results from section 2.6: (will be shown in detail later in
the lecture)

( )
2
1 1 2
12 1 2
V V V
S
Z Z
= +

Z
V
1
V
2
I
12
S
12
-S
12
S
21
-S
21
+ +
-
-

9
( )
2
1 1 2
12 1 2
cos cos
V V V
P
Z Z
= +

( )
2
1 1 2
12 1 2
sin sin
V V V
Q
Z Z
= +

where
1 1 1
V V = and
2 2 2
V V = and the impedance is Z Z = . The angle
defined as
1 2
= is often called the "power angle". N.B. This is very different from
the "power factor angle" discussed earlier.

If we assume that the circuit above represents two generators connected by a transmission
line, then the equations above still apply. In particular the impedance of a transmission
line may be assumed purely inductive, i.e. 0 Z jX = + . In this case the angle 90 =
!
and
the equations above simplify to:
( )
1 2
12 1 2
sin
V V
P
X
=


( )
1
12 1 2 1 2
cos
V
Q V V
X
=



Since the line is "lossless" then power consumed by the line is zero, i.e.
12 21
0 P P + = . In
terms of the power angle, these equations become:


1 2
12
sin
V V
P
X
=


1
12 1 2
cos
V
Q V V
X
=



Observations (assuming 0 R ):

1. Usually is very small (less than 10 degrees), thus
12
sin P , i.e. small changes in
power angle greatly change the real power flow and not the reactive power. If
1 2
> then power flows from node one to node two. If
1 2
< , then power flows in
the opposite direction (from node 2 to node 1).
2. Maximum power transfer occurs when 90 =
!
and is given by:
1 2
max
V V
P
X
= .

10
3. Since 0 ,
1 2
Q V V . Thus small changes in
1 2
V V greatly affect Q but not
P .

Thus, to control real power, we need to change the power angle . This is done be
increasing prime mover power (mechanical power driving the generator). To control
reactive power, we need to change the difference in voltage magnitude. This is done by
changing the DC excitation of one generator.

Example 2.5

Assume
1
120 5 V = and
2
100 0 V = . Let 1 7 Z j = + . Determine the real and
reactive power supplied or received by each source and the power loss in the line.

The Matlab code follows:

R = 1; X = 7; Z = R +j*X;
V1 = 120*(cos(-5*pi/180) + j*sin(-5*pi/180));
V2 = 100+j*0;
I12 = (V1 - V2)/Z, I21 = -I12;
S12 = V1*conj(I12), S21 = V2*conj(I21)
SL = S12 + S21
PL = R*abs(I12)^2, QL = X*abs(I12)^2

I12 =
-1.0733 - 2.9452i
S12 =
-9.7508e+001 +3.6331e+002i
S21 =
1.0733e+002 -2.9452e+002i
SL =
9.8265 +68.7858i
PL =
9.8265
QL =
68.7858


Example 2.6:

Repeat example 2.5 using a Matlab program such that the angle of source 1 is changed
from -30 to 30 degrees in steps of 5 degrees each.

We cannot execute "chp2ex6" from within the notebook because input values are to be
entered by the user. Hence we go to a Matlab window and execute this command.

GO TO A MATLAB WINDOW AND EXECUTE "chp2ex6".


1
Review of Chapter 2, Plus Matlab Examples (Part II)


2.7 Balanced Three-Phase Circuits

Let 1 120 a =
o
. Clearly then
2 *
a a = and
3
1 a = . The triple (1, a ,
2
a ) form a set of balanced phasors, thus
2
1 0 a a + + = . This triple
represents the three cube roots of 1.


Y-connected generator, positive sequence:

0
An p
E E =

2
120
Bn p An
E E a E = =
240
Cn p An
E E aE = =

Here
p
E is the "phase" voltage. Note that if we know
An
E we can
find the other two voltages in terms of
An
E .

It is noted that line-to-line voltage
L
V is larger than the phase voltage by a factor of 3
and is ahead by 30
o
, i.e. 3 30
L L p
E E

=
o
. However, line current and phase current
are identical, i.e.
L p
I I = .


Y-connected generator, negative sequence:

0
An p
E E =
120
Bn p An
E E aE = =

2
120
Cn p An
E E a E = =

The negative sequence connection is obtained by "interchanging" any
two phases of the positive sequence connection. Power systems use only positive
sequence connected generators. Also, every effort is made to maintain balanced voltages
and loads.



1
a
a
2
E
An
E
Bn
E
Cn

t
E
An
E
Cn
E
Bn

t
2
V
an
n
V
bn
V
cn
V
ab
V
bc
V
ca
2.8 Y-Connected Loads

0
An p
V V =

2
120
Bn p An
V V a V = =
240
Cn p An
V V aV = =

The line-to-line voltages are found as follows:

( ) 1 0 1 120 3 30
ab an bn p p
V V V V V = = =

Similarly we find:


3 90
bc p
V V =
3 150
ca p
V V =




Note: all these results apply provided the
reference phasor is as shown. This is
arbitrarily chosen, for example above, it is
assumed that V
an
is the reference vector (at
zero degrees). If a different reference is
chosen, the diagram stays the same though it
will be rotated.

2.9 Delta Connected Loads

In a similar way we see that
L p
V V = (line voltage and phase voltage are the same for
delta connected loads) and

3 30
a p
I I =
3 150
b p
I I =
3 90
c p
I I =
and in general:
3 30
L p
I I =
V
an
V
bn
V
cn
V
ab
V
bc
V
ca
3
2.10 -Y Transformation for Balanced Loads

3 / 3
Y Y
Z Z Z Z

= =

2.11 Per-Phase Analysis

1. It is assumed that everything is "balanced".
2. Change all delta connected loads/sources to Y connections, this provides a neutral
point.
3. All the neutral points are at the same potential, hence all the neutral points may be
connected
4. This breaks up the circuit into three entirely separate circuits, one for each phase.
5. Solve the single-phase circuit for phase "a". The other phases are the same after a
phase shift of 120
o
.


2.12 Balanced Three-Phase Power

Recall that single phase instantaneous power was sinusoidal and had a frequency twice
that of the line. Its average value was called the real power. Here is the single phase
power equation:

( ) ( ) ( ) 1 cos2 sin 2
v v
p t P t Q t = + + + +

**Single phase**

The three-phase instantaneous power equation will be derived below.

Suppose the voltages of a three-phase supply are given by:

( ) 2 cos
an p v
v V t = +

( )
2 cos 120
bn p v
v V t = +
o


( )
2 cos 240
cn p v
v V t = +
o


Assume the currents are given by:

( ) 2 cos
a p i
i I t = +

( )
2 cos 120
b p i
i I t = +
o


( )
2 cos 240
c p i
i I t = +
o


The instantaneous power is then found by adding the powers for the three phases:

4

( )
3 an a bn b cn c
p t v i v i v i

= + +
Thus we have:


( ) ( ) ( )
( ) ( )
( ) ( )
3
2 cos cos
2 cos 120 cos 120
2 cos 240 cos 240
p p v i
p p v i
p p v i
p t V I t t
V I t t
V I t t




= + +
+ + +
+ + +


Using the identity ( ) ( )
1 1
cos cos cos cos
2 2
A B A B A B = + + we have:


( ) ( ) ( )
( ) ( )
( ) ( )
3
cos cos 2
cos cos 2 240
cos cos 2 480
p p v i v i
p p v i v i
p p v i v i
p t V I t
V I t
V I t




= + + +

+ + + +

+ + + +



Note that the terms involving time are a balanced set hence add to zero and we have:

( ) ( )
3
3 cos
p p v i
p t V I

=

If we let
v i
= (the impedance angle) then:

( )
3 3
3 cos
p p
p t P V I

= =

It is interesting to note that the RHS is independent of time! Each
phase by itself has a pulsating power, but when the three are added,
the result is a constant power compare with single phase!

The concept of complex power is now extended to three phase, thus we define


3
3 sin
p p
Q V I

=


3 3 3
S P jQ

= +

In terms of phase voltages and currents, we also have:


*
3
3
p p
S V I

=

5
Since in the Y-connection we have
3
L
p
V
V = and
p L
I I = , then we have:


( )
3 3
3 cos
3 cos
p p
L L
p t P V I
V I

= =
=



3
3 sin
L L
Q V I

=

In case of the -connection
p L
V V = and
3
L
p
I
I = giving rise to the same equations
above. The two equations above are valid for - as well as Y-connected circuits: They
involve line-to-line quantities.

Very Important: The angle in the power equations above is the angle between the
phase voltage and phase current (not line voltages nor line currents). Failure to
recognize this fact can lead to serious computational errors.

Example 2.7

A three phase line has an impedance of 2 4 j + . The line-to-line voltage is 207.85 V.
Two three-phase loads are connected to this line in parallel: one is a -load, the other is a
Y-load. Each leg of the -load has an impedance of 60 45 j , and each leg of the Y-
load has an impedance of 30 40 j + . Taking the phase voltage
a
V as a reference,
determine:

(a) The current, real power, and reactive power drawn from the supply.
(b) The line voltage at the combined loads.
(c) The current per phase in each load.
(d) The total real and reactive powers in each load and the line.

The Matlab program follows:

V1 = 207.85/sqrt(3);
ZL = 2 + j*4; Z1 = 30 + j*40; % Note ZL is
BETTER than Zl, which looks like Z1
Z2 = (60-j*45)/3; % Z2 is Z/3
Z = ZL + Z1*Z2/(Z1+Z2);
I=V1/Z, S=3*V1*conj(I)
V2 = V1- ZL*I
V2ab =sqrt(3)*(cos(pi/6)+j*sin(pi/6))*V2
I1=V2/Z1, I2=V2/Z2
Iab=I2/(sqrt(3)*(cos(pi/6)-j*sin(pi/6)))
S1=3*V2*conj(I1), S2=3*V2*conj(I2)
SL = 3*ZL*abs(I)^2
Stotal=S1+SL+S2
V
1
V
2
Z
L
Z
Y
Z

/3
I I
1
I
2
+
+
- -
6

I =
5.0001
S =
1.8001e+003
V2 =
1.1000e+002 -2.0000e+001i
V2ab =
1.8232e+002 +6.5264e+001i
I1 =
1.0000 - 2.0000i
I2 =
4.0001 + 2.0000i
Iab =
1.4227 + 2.1547i
S1 =
4.5002e+002 +6.0002e+002i
S2 =
1.2000e+003 -9.0003e+002i
SL =
1.5001e+002 +3.0001e+002i
Stotal =
1.8001e+003


Example 2.8

A three-phase line has impedance of 0.4 2.7 j + per phase. The line feeds two balanced
three phase loads that are connected in parallel. The first load is absorbing 560.1 kVA at
0.707 power factor lagging. The second load absorbs 132 kW at unity power factor. The
line-to-line voltage at the load end of the line is 3810.5 V. Determine:

(a) The magnitude of the line voltage at the source end of the line.
(b) Total real and reactive power losses in the line.
(c) Real and reactive power supplied at the sending end of the line.

The Matlab program follows:

V2 = 3810.5/sqrt(3);
Zl = 0.4 +j*2.7;
S1 = 560.1*(cos(pi/4) + j*sin(pi/4)); S2 = 132;
SR = S1+ S2
I = conj(SR)*1000/(3*conj(V2))
V1 = V2 + Zl*I
V1L = sqrt(3)*abs(V1)
SL = 3*Zl*abs(I)^2/1000
SS = 3*V1*conj(I)/1000
SS = SR + SL

SR =
5.2805e+002 +3.9605e+002i
I =
80.0079 -60.0078i
V1 =
7
2.3940e+003 +1.9202e+002i
V1L =
4.1599e+003
SL =
12.0026 +81.0179i
SS =
5.4005e+002 +4.7707e+002i
SS =
5.4005e+002 +4.7707e+002i



Note on units. In reality, P, Q, and S, all have unit of Watt, but to distinguish them apart
we assign units as follows:


S=VI*=P+jQ P Q |S|=|V||I|
NAME Complex Power Real Power Reactive Power Apparent Power
unit W+jvar or VA at
power factor
W, kW, MW var, kvar, Mvar VA, kVA, MVA




Chapter 3. Generator and Transformer Models;
The Per-Unit System

3.1 Introduction

1. Use of the "per phase" basis to represent a three-phase balanced system.
2. Use the model to describe transmission lines (see Chapter 5).
3. Simple models for generators and transformers to study steady state balanced
operation.
4. One-line diagrams to represent a three phase system.
5. The per-unit system and the impedance diagram on a common MVA base.


3.2 Synchronous Generators

Note the coils aa', bb', and cc' are 120
o
apart. Axis of a-coil is the x-axis as shown.
These are
"concentrated" full
pitch windings. In
real machines the
windings are
distributed among
many slots, and often
are not full pitch.
We assume the
windings produce a
sinusoidal mmf
around the rotor
periphery (angle
around the airgap
with respect to
winding axis) .
Assume the rotor is
excited with DC
current
f
I producing
a flux which
rotates with the rotor
at speed . At time t the rotor would have moved an angle t .
The flux linkage for an N-turn a-winding will be maximum
( )
N at 0 t . The
flux linkage with the a-winding will go to 0 at / 2 t . The flux linkage will go to
( )
N at t , and back to 0 at 3 / 2 t . When 2 t the flux linkage will
be back to
( )
N . Thus a whole cycle This repeats every revolution of the rotor.
a
a'
b
b'
c
c'
r r'
m
n
F
s
F
r
F
sr

Rotor position shown at wt = p/2


2
The flux linkage with winding "a" is thus given by: cos
a
N t
Using Faraday's law, the voltage induced in phase "a" is given by:


max
max
sin
sin
cos
2
a
d
e N t
dt
E t
E t

_


,


Where
max
2 E N fN , since 2 / 2 4.44 , the rms value of the generated
voltage is given by:

4.44 E fN


The frequency is a
function of speed and the
number of poles, thus:
2 60
P n
f where n is
the speed in rpm (the
synchronous speed) and
P is the number of poles
(always an even
number).

If the phase "a" is
connected to a load, then
a current
a
i will flow.
Depending on the load,
this current will have a
phase angle, say (see
figure) lagging the
generated voltage
a
e
which is along the x-
axis. Again, this is
shown in the figure as the line "mn". N.B. The line "mn" is attached to and rotates with
the rotor. Be careful also that the "Flux" vectors are spatial while the voltage and current
vectors are phasors (in time).

The same is true for phases "b" and "c" but they will "lag" the voltage in phase "a" by
120 and 240 degrees respectively.

a
a'
b
b'
c
c'
r r'
m
n
F
s
F
r
F
sr

Rotor position shown at wt = p/2


3
Since
( )
sin
a
e t , then we have:

( )
( )
( )
max
max
max
sin
sin 120
sin 240
a
b
c
i I t
i I t
i I t






o
o


Since mmf is proportional to the current, we then have:


( )
( )
( )
sin
sin 120
sin 240
a m
b m
c m
F F t
F F t
F F t






o
o


We now take components of these phasors along the line "mn" and in quadrature with it.

Along "mn" we have:


( ) ( ) ( ) ( )
( ) ( )
1
sin cos sin 120 cos 120
sin 240 cos 240
m m
m
F F t t F t t
F t t


+ +

o o
o o


Using the identity ( )
1
sin cos sin 2
2
the above equation becomes:

( ) ( ) ( )
1
sin 2 sin 2 120 sin2 240
2
m
F
F t t t
1
+ +
]
o o


It is noted that this expression is the sum of three balanced phasors hence is equal to zero.

Next we consider the components of the mmf perpendicular to "mn":


( ) ( ) ( ) ( )
( ) ( )
2
sin sin sin 120 sin 120
sin 240 sin 240
m m
m
F F t t F t t
F t t


+ +

o o
o o


Using the identity ( )
2
1
sin 1 cos2
2
we have:

( ) ( ) ( ) { }
2
3 cos2 cos2 120 cos2 240
2
m
F
F t t t
1
+ +
]
o o


The three cosine terms add to zero (balanced phasors) thus we have:

4

3
2
S m
F F

Thus the result of having three pulsating single phase fluxes produce (when applied
symmetrically) a constant flux perpendicular to line "mn" and rotates at the same speed
n as the rotor. This flux is called the "armature reaction" to the field of the rotor
r
F . the
various fields are shown for one phase (say phase "a") in the diagram below: Note that F
s

is perpendicular to line "mn" and rotates with it at the same speed. The fact that F
s

(which is proportional to I
a
) is perpendicular to line "mn" indicates that that we can
theorize that the armature current I
a
is producing a reactive voltage drop parallel to line
"mn" due to inductance in the machine.

Run the command "rotfield" in a Matlab window to see a demo of the rotating field.

is the angle between and
a
V I it is the power factor angle
is the angle between and E V it is the power angle

First the rotor field F
r
produces the no-load generated voltage E (at zero armature
current). Note that E lags F
r
by 90 degrees. E is called the "excitation" voltage. It is
directly proportional to the field current.

The voltage/current phasors for phase "a" are shown above lagging the flux diagram by
90 deg. Note the above diagram is a "hybrid" combining spatial and temporal vectors.

E
F
r
F
sr
F
s
I
a
m
n
V
R
a
I
a
jX
l
I
a
jX
ar
I
a
E
sr
E
ar
Figure 3.2 Page 53
5
Second, assume the armature carries a current to a load, now the armature reaction flux F
s

is produced. This is perpendicular to line "mn". The two fluxes (due to rotor and
armature) combine together to form the "resultant" flux F
sr
. The resultant flux induces
the generated on-load emf E
sr
. The armature mmf F
s
induces the voltage E
ar
known as
the armature reaction voltage.

In all cases, each mmf produces a voltage lagging the mmf by 90 degrees.

Note that the voltage E
ar
leads F
s
(hence I
a
) by 90 degrees. Thus we can theorize an
inductor model for this relationship with reactance X
ar
i.e.
ar ar a
E jX I . X
ar
is known as
the reactance of armature reaction. Thus we now have the circuit equation:


sr ar a
E E jX I +

The terminal voltage V is found by considering the armature resistance and leakage
reactance thus:
( )
a ar a
E V R j X X I + + + 1
] l


Which can be simplified to:


[ ]
a s a
E V R jX I + +

where
s ar
X X X +
l
is known as the synchronous reactance.

The angle between V and I
a
is the power factor angle. The power angle is the angle
between E and V called . The circuit model of the machine is as shown below:

The resistance R
a
is much smaller than
the synchronous reactance X
s
and is often
neglected. This is shown in the simplified
one line diagram below where the
machine is shown connected to an infinite
bus. An infinite bus has a voltage V (usually assumed at an angle zero) which does not
change no matter how the generator is controlled or excited.

A figure of merit of a generator is the percent voltage regulation VR defined as:

100 100
nl rated rated
rated rated
V V E V
VR
V V


+
-
Load
E
R
a
jX
s
+
-
V
I
a
jX
s
E
V
I
a
6

The smaller the VR the better. In a real generator, the regulation is a few percent.


3.3 Steady-State Characteristics, Cylindrical Rotor

Power factor control. Most generators are connected to a large power grid. This is an
infinite bus (its voltage, angle and frequency are constant). Assuming the generator has
small leakage reactance and small armature resistance, then its model is shown below:

We have two ways of calculating the power (per
phase):

1.
1
cos
a
P V I


2.
1
sin
s
E V
P
X



Equating these two equations we have sin cos
s a
E X I . Assuming the power is
constant, then from equation 1 we have: cos .
a
I const This locus is shown as the
vertical dashed line in the figure below.

From equation 2, assuming the power is a
constant we have: sin . E const This locus is
the horizontal dashed line shown.

Note that for this load, the minimum armature
current is
2 a
I when the power factor of the
generator is unity. Note that
2
E is directly above
the voltage V . Thus
2 a
I and V are in phase,
producing unity power factor. If the excitation is
increased, the emf of the generator is increased to
some value say
1
E as shown. Clearly the voltage of the generator leads the current hence
it is like an inductor consuming vars (lagging power factor). On the other hand, if the
excitation is reduced below that for unity power factor, the emf of the generator is
smaller, say
3
E which lags
3 a
I . Now the power factor of the generator is "leading", i.e.
it is like a capacitor (current leading voltage). A plot of the variation of
a
I as a function
of field current is known as a V-curve. A family of such curves, one at each load is
known as the V-curves of the generator. The minimum point on each curve corresponds
to unity power factor. To the right (higher excitation) is the lagging power factor area
and to the left is the leading power factor zone.

jX
s
E
V
I
a
I
a1
I
a2
I
a3
V
E
1
E
2
E
3

1
q
1
7
Power angle characteristics. Consider the machine connected to an infinite bus and
neglecting armature resistance, the machine model is shown below:

*
3
3
a
S VI


But we have
0
a
s
E V
I
Z

thus for complex power:


2
3
3 3
s s
E V V
S
Z Z

thus the real and imaginary powers are given by:



( )
2
3
3 cos 3 cos
s s
E V V
P
Z Z


( )
2
3
3 sin 3 sin
s s
E V V
Q
Z Z



and if we neglect the armature resistance we have: (Note that 90
o
and
s s
Z X )

3
3 sin
s
E V
P
X

and
( )
3
3 cos
s
V
Q E V
X

. Thus the real power is a function of


the power angle and varies as sin . The maximum power is given by:
max(3 )
3
s
E V
P
X

.
The angle starts at zero when the machine does not deliver any power. As the
machine is loaded, the power delivered increases and the angle increases. As the angle
reaches 90
o
, maximum power is reached. If prime mover power is increased beyond
this point, the machine will speed up and loose synchronism. This means the tie between
the machine and the power grid can handle only the maximum power shown above. In
real operation, the power angle is very small of the order of 10 degrees and less. The
power angle can "swing" by larger amounts under transient conditions.

The equation for Q shows that
( )
3
3
s
V
Q E V
X

; . Thus if E V > the generator


delivers reactive power and the generator is said to be overexcited. If E V < , the
generator is underexcited and the reactive power delivered to the bus is negative, thus the
generator is receiving reactive power from the bus. Generators are usually operated in
the overexcited mode since they need to supply the system with reactive power: most our
domestic loads are reactive.


Example 3.1

jX
s
E
V
I
a
8
A 50-MVA, 30-kV, three phase, 60 Hz synchronous generator has Xs = 9 per phase
and a negligible resistance. The generator is delivering rated power at 0.8 power factor
lagging at the rated terminal voltage to an infinite bus.

(a) Determine the excitation voltage per phase E and the power angle .
(b) With the excitation held constant at the value found in (a), the driving torque is
reduced until the generator is delivering 25 MW. Determine the current and the
power factor.
(c) If the generator is operating at the excitation voltage of part (a), what is the steady-
state maximum power the machine can deliver before losing synchronism? Also, find
the armature current corresponding to this maximum power.

The Matlab program follows:

Ra=0; Xs = 9; Zs = Ra + j*Xs;
S = 40 + j*30; V = 30/sqrt(3); % MW and kV
Ia1 = conj(S)*1000/(3*conj(V)); % Amp
Ia1M = abs(Ia1), Ia1ang=angle(Ia1)*180/pi
E1 = V + Zs*Ia1*0.001; % kV
E1M = abs(E1), delta1 = angle(E1)*180/pi % kV, deg
disp('(b)')
P = 25; % MW
delta2 = asin(P*Xs/(3*abs(E1)*V)); delta2d = delta2*180/pi
E2 = E1M*(cos(delta2) +j*sin(delta2)); % kV
Ia2 = (E2 - V)*1000/Zs; Ia2M = abs(Ia2), Ia2ang=angle(Ia2)*180/pi
PF = cos(angle(Ia2))
disp('(c)')
Pmax = 3 *E1M*V/Xs % kW
E3 = E1M*(cos(pi/2) +j*sin(pi/2)); % kV
Ia3 = (E3 -V)*1000/Zs; % A
Ia3M = abs(Ia3), Ia3ang=angle(Ia3)*180/pi
PF = cos(angle(Ia3))

Ia1M =
962.2504
Ia1ang =
-36.8699
E1M =
23.5584
delta1 =
17.1027
(b)
delta2d =
10.5914
Ia2M =
807.4918
Ia2ang =
-53.4284
PF =
0.5958
(c)
Pmax =
136.0147
9
Ia3M =
3.2489e+003
Ia3ang =
36.3239
PF =
0.8057


Example 3.2

The generator of example 3.1 is delivering 40 MW at a terminal voltage of 30 kV.
Compute the power angle, armature current, and power factor when the field current is
adjusted for the following excitations:

(a) The excitation voltage is decreased to 79.2 percent of the value found in example 3.1.
(b) The excitation voltage is decreased to 59.27 percent of the value found in example
3.1.
(c) Find the minimum excitation below which the generator will lose synchronism.

The Matlab program follows:

Ra=0; Xs = 9; Zs = Ra + j*Xs;
V = 30/sqrt(3); P = 40;
E1M = 0.792*23.558;
delta1 = asin(P*Xs/(3*abs(E1M)*V)); delta1d = delta1*180/pi
E1 = E1M*(cos(delta1) +j*sin(delta1));
Ia1 = (E1 - V)*1000/Zs; Ia1M = abs(Ia1), Ia1ang=angle(Ia1)*180/pi
PF = cos(angle(V/Ia1))
disp('(b)')
E2M = 0.5925*23.558;
delta2 = asin(P*Xs/(3*E2M*V)); delta2d = delta2*180/pi
E2 = E2M*(cos(delta2) +j*sin(delta2));
Ia2 = (E2 - V)*1000/Zs; Ia2M = abs(Ia2), Ia2ang=angle(Ia2)*180/pi
PF = cos(angle(Ia2))
disp('(c)')
E3M = P *Xs/(3*V*1)
E3=E3M*(cos(pi/2)+j*sin(pi/2));
Ia3 = (E3 -V)*1000/Zs;
Ia3M = abs(Ia3), Ia3ang=angle(Ia3)*180/pi
PF = cos(angle(Ia3))

delta1d =
21.7975
Ia1M =
769.8005
Ia1ang =
-0.0283
PF =
1.0000
(b)
delta2d =
29.7592
Ia2M =
962.7210
10
Ia2ang =
36.9072
PF =
0.7996
(c)
E3M =
6.9282
Ia3M =
2.0728e+003
Ia3ang =
68.1986
PF =
0.3714

Notes: in (a) the generator is operating at unity power factor. In both (b) and (c) the
generator is under-erexcited and is receiving reactive power (power factor of load is
leading, i.e. capacitive load and the generator is inductive.) In the real world the load is
usually inductive (lagging power factor) and the generator is over-excited, delivering
reactive power.

Example 3.3

For the generator of example 3.1, construct the v-curve for the rated power of 40 MW
with varying field excitation from 0.4 power factor leading to 0.4 powr factor lagging.
Assume the open-circuit characteristic in the operating reagion is given by E = 2000I
f
.

The Matlab program follows:

P = 40; % real power, MW
V = 30/sqrt(3)+ j*0; % phase voltage, kV
Zs = j*9; % synchronous impedance
ang = acos(0.4);
theta=ang:-0.01:-ang; % Angle from 0.4 leading to 0.4 lagging pf
P= P*ones(1,length(theta)); % generates P array of same size
Iam = P./(3*abs(V)*cos(theta)); % current magnitude kA
Ia = Iam.*(cos(theta) + j*sin(theta)); % current phasor
E = V + Zs.*Ia; % excitation voltage phasor
Em = abs(E); % excitation voltage magnitude, kV
If = Em*1000/2000; % field current, A
plot(If, Iam), grid
xlabel('If, A'), ylabel('Ia, kA')
text(3.4, 1, 'Leading pf'), text(13, 1, 'Lagging pf')
text(9, .71, 'Upf')

11
2 4 6 8 10 12 14 16 18
0.6
0.8
1
1.2
1.4
1.6
1.8
2
If, A
I
a
,

k
A
Leading pf Lagging pf
Upf



Note that the over-excited generator (load is inductive, generator is capacitive) is the
normal way to operate the generator since most loads are inductive. If the load becomes
capacitive, then the generator is operated in the under-excited mode.


Chapter 3. Generator and Transformer Models;
The Per-Unit System (Part II)

3.4 Salient-Pole Synchronous Generators

Due to the non uniform airgap of the salient poles, the flux is stronger in the direct axis
than in the quadrature axis. Thus the reactance
d
X in the direct axis direction is larger
than
q
X in the quadrature axis direction. The common way of dealing with this is to
decompose the armature current into two components: one along the direct axis and one
along the quadrature axis. Now the voltage can be found by adding the components
along these directions. It is noted that an Ampere in the direct axis direction will produce
more voltage than an Ampere in the quadrature direction (due to the difference in
magnitude between
d
X and
q
X ). This is illustrated in the figure below.

Note how the contribution to voltage in the direct axis is much larger than in the
quadrature axis. Note also that the arma ture resistance is neglected. It is no longer
possible to represent the machine by a simple equivalent circuit.

The excitation voltage is given by:

cos
d d
E V X I +

The three phase power at the generator is given by:

3 cos
a
P V I

Using the figure above we see that


cos
cos sin
a
q d
I ab de
I I


+
+

Therefore the power is:
( )
3 cos sin
q d
P V I I + . From the diagram we see that
sin
q q
V X I or
sin
q
q
V
I
X

. Also from an equation above we have:


cos
d
d
E V
I
X

, therefore the power becomes
sin cos
3 cos sin
q d
V E V
P V
X X


_

+


,
. Note that
sin2
sin cos
2

, thus:
a
b
c
d
e
E
I
d
I
q
I
a
V
jX
q
I
q
jX
d
I
d


2

2 2
sin 3 3
si n2
3
2
d q d
E V V V
P
X X X


1
+ 1
1
]


Which simplifies to the equation in the book (3.31) page 64:


2
sin
3 3 sin 2
2
d q
d d q
X X E V
P V
X X X

+

Remarks: the author does not show the details shown above, but the results are well
known. Also note that if the machine is cylindrical, then
d s
X X and
q d s
X X X ,
thus the power equation becomes the normal power, namely:
sin
3
s
E V
P
X

. Also
note that the new term known as the reluctance power is at frequency (in space) twice
that of the normal power. Thus the power-angle relationship is changed from a perfect
sinusoid. This is shown below:

delta=0:.001:pi;
Xs=1; Xd=1; Xq=0.6; E=1; V=0.7;
P=3*E*V*sin(delta)/Xs;
P2=3*V^2*(Xd-Xq)*sin(2*delta)/(2*Xd*Xq);
Psal=P+3*V^2*(Xd-Xq)*sin(2*delta)/(2*Xd*Xq);
plot(delta,P,delta,Psal,delta,P2),grid

0 0.5 1 1.5 2 2.5 3 3.5
-0.5
0
0.5
1
1.5
2
2.5


3

Note that the salient-pole machine has greater peak power (stiffer), and this peak power
occurs at an angle slightly less than / 2 radians.

3.6 Equivalent Circuit of a Transformer

First consider an ideal transformer, one with no leakage and infinite permeability. Let the
transformer have N
1
turns on the left side and N
2
turns on the right side. Assume the
current IN on the left is
2
I and OUT on the right is I
2
. In such a case we have (assuming
sinusoidal flux cos t ):

( )
1 1
1
1max
sin
cos 90
d
e N
dt
N t
E t


+
o

where
1max 1
2 E fN . Thus the RMS value of E
1
is
1 1
4.44 E fN . Since the same
flux links both windings, we have on the right:
2 2
4.44 E fN . Also, since the core has
no losses, we must have
2 1 2 2
I N I N . This leads to the relationships:

1 2 1
2 2 2
E I N
E I N


I.e. in the ideal transformer, the ratio of currents and voltages is the turns ratio (though
the current ratio is inversely so). The ideal transformer is shown below:

In a real transformer, there would be some leakage
flux, also some losses. There is also a magnetizing
current. These are simulated by adding components
to the ideal model, thus a real transformer would be as
shown below, this is the T-equivalent model or
equivalent circuit.

Assuming the load on an ideal transformer has an impedance
2
Z , then we have:
2
2
2
E
Z
I
.
I
'
2

1

2

2
The ideal tranformer
I
'
2

1

2

2
Z
1
=R
1
+jX
1
Z
2
=R
2
+jX
2
V
1
+
-
V
2
+
-
R
c1
jX
m1
I
0
I
m
I
c

1
N
1
:N
2
Equivalent circuit of a power transformer

4
The impedance seen at the left of the ideal transformer is
1
2 2
2
1 1 2 1 2
1 2
2
2 2 2 2
2
1
N
E
E N E N N
Z Z
N
I N I N
I
N
_ _

, ,
. Thus impedances are transformed as the turns
ratio squared (or the inverse turns ratio squared going the other way). Let
1
2
N
N
.
Then:
2 2 2 2
E V Z I + . But,
1 2
E E and
2 2
I I , therefore:

2
1 2 2 2
2 2 2
E V Z I
V Z I
+

+

where
2 2
2 2 2 2 2
Z R jX R X + + .
Using these transformations, the model of the transformer will be as shown in the figure
below:


This is then approximated with the figure below:



It is easier to perform computations with the latter figure. Note that this approximate
equivalent model can be either on the left side (primary) as shown above, or the right side
(secondary) as shown below.

V'
1
+
-
R
c2
jX
m2
I'
0
I'
m
I'
c

1
Z
e2
=R
e2
+jX
e2
I
2
V
2
+
-


1
Z
1
=R
1
+jX
1
V
1
+
-
R
c1
jX
m1
I
0
I
m
I
c

1
+
-
Z'
2
=R'
2
+jX'
2
I'
2
=I
2
N
2
/N
1
V'
2
=V
2
N
1
/N
2
+
-
V
1
+
-
R
c1
jX
m1
I
0
I
m
I
c

1
Z'
e1
=R'
e1
+jX'
e1
I'
2
=I
2
N
2
/N
1
V'
2
=V
2
N
1
/N
2
+
-

5
Furthermore, if we neglect the shunt parts (which are relatively of high impedance) the
result is only an RL equivalent as shown below. In most applications, only the reactance
X is kept since R is so much smaller it is neglected.

V
1
+
-

1
Z
e1
=R
e1
+jX
e1
+
-
V'
2
=V
2
N
1
/N
2

V'
1
=V
1
N
2
/N
1
+
-

2
Z
e2
=R
e2
+jX
e2
V
2
+
-


Simplified models referred to one side.


3.7 Determination of Equivalent Circuit Parameters

The OC and SC tests are used to determine the equivalent circuit parameters. This is
done using the approximate model of figure 3.12 (top of page 68). In the OC test, one
side is excited to rated voltage
1
V (usually the low voltage side) and the other side is open
circuited. The input power
0
P and current
0
I are measured.
Since the current is small compared to rated value, the losses
in the series arms of the model may be neglected and it is
assumed that the power lost is the "core losses" (the iron core
losses) in the shunt components
1 c
R and
1 m
X which are in
parallel. The computations required are as follows:
2
1
1
0
c
V
R
P
,
1
1
c
c
V
I
R
,
2 2
0 m c
I I I and
1
1 m
m
V
X
I
.

In the SC test a reduced voltage
SC
V is applied to one
side while the other side is shorted. The input current
SC
I and input power
SC
P are measured. In this test the
applied voltage is raised till the current on the short
circuit side reaches rated value. Note that only a few
percent of rated voltage is needed to produce rated
currents under this SC condition.

Since the currents are so high and the voltage applied is
so low, the shunt branch may be neglected, and the series branch is where the losses are
(approximately). Note that the series branch is the "equivalent" values of the model
1 1 1 e e e
Z R jX + (see figure 3.15, top of page 70). The equations needed to compute these
parameters are:
1
SC
e
SC
V
Z
I
,
1
2
SC
e
SC
P
R
I
and
2 2
1 1 1 e e e
X Z R .
V
1
+
-
R
c1
jX
m1
I
0
I
m
I
c
I
0
OC Test
Z
e1
= R
e1
+jX
e1
V
SC
+
_
I
SC
Short Circuit Test

6


3.8 Transformer Performance

If we have the approxi mate equivalent model from the OC and SC tests, we can measure
transformer performance. The efficiency of these transformers tends to be high (above
95% efficient). The efficiency is computed as follows:

( )
2
cu c
power out n S PF
power in n S PF n P P



+ +

where n is the fraction of the full-load power, S is the rated apparent power or nominal
power in volt-ampere, and
cu
P is the full load copper loss. For a three phase transformer
these are given by:
2 2
3 S V I and
2
2 2
3
cu e
P R I . Also,
c
P is the iron loss at rated
voltage. As the book says, "it can be easily shown that maximum efficiency occurs when
copper losses equal iron losses at n per-unit loading" see page 71!

At any rate, we do not need these details. There is also a program called "trans" which
we have seen before that solves these equations for efficiency. Other important equations
are those for percent regulation. These are referred to either side and have to do with
voltage drop at full load compared to no load voltage. See equations on bottom of page
71.

Example 3.4

Run Trans with the given data! see page 72.


3.9 Three-Phase Transformer Connections

Y Y : This is rarely used due to third harmonics. Often a third "tertiary" winding is
added connected in to provide a path for the third harmonic current thus allowing the
voltage to remain sinusoidal and almost distortion free. Voltage out is in phase with
voltage in.

: This allows for the third harmonic, it lacks the neutral connection. It has the
advantage that one bank (assuming three banks are used) of transformers can be removed
for repairs, and it will continue to operate (at reduced power) till the repair is completed.
It is then connected as a V V or simply V-connection.

Y : Is used frequently to raise voltage. Note that there is a connection voltage gain (in
addition to the inherent turns ratio gain). There is also a phase shift of thirty degrees.
The neutral is often grounded.

Y : Is used to lower the voltage from high transmission values. Note that it also
compensates for phase shifts on a Y connection thus bringing the line back in phase.

7


3.9.1 The Per-Phase Model of a Three-Phase Transformer

Usually the low voltage side is designated as the X side and the high voltage side as the H
side. Thus
XP
V refers to the phase voltage on the low side. In the case of and
Y Y connections the phase does not change going from one side to the other. Also, the
voltage gain is essentially the turns ratio. This is not the same for a Y or a Y
connection. Take for example the Y connection. First of all there is a 30
o
phase
shift. It is the standard (American Standards Association) to make the connection such
that the line voltage on the HV side leads the corresponding line voltage on the LV side
by 30
o
regardless of which side the Y or the is on. Second, the Y-side has a
connection gain of 3 . Assume the Y-side is the HV side, thus we have for a Y
connection: (let
H HP
X XP
N V
a
N V
)

3
HL HP
XL XP
V V
V V


thus the ratio of the line-to-line voltages of a Y assuming the Y is the high voltage
side is 3 3
HL HP
XL XP
V V
a
V V
. I.e. the total gain is the connection gain multiplied by the
turns ratio.

When dealing with a Y or a Y connection, it is best to transform the delta to a Y
then use a per-phase diagram. See figure 3.19 page 77.


3.10 Autotransformers

When the two windings of a transformer are connected together it becomes an
autotransformer. Note that the two sides are now no longer isolated from each other as in
a conventional transformer. This may be undesirable. If
the isolation is not an important feature, then this
connection has several advantages as we show below. The
two winding transformer is shown to the right. Below that
we show the two sides connected into an autotransformer.
Note that now the two sides are no longer isolated.

The winding voltages and currents are related in the same
manner for the autotransformer as for the two-winding
transformer, thus:
1 1
2 2
V N
a
V N
and
2 1
1 2
I N
a
I N
. From the
N
1
:N
2
V
1
V
2
+
+
_
_
X
2
X
1
H
1
H
2
I
1
I
2
V
L
+
_
H
1
H
2
I
2
V
1
+
_
X
1
I
1
I
H
X
2
V
2
V
H
I
L

8
figure we have: ( )
1
2 1 2 2
2
1
H L
N
V V V V V a V
N
+ + + . Thus we have 1
H
L
V
a
V
+ . Since
we have the ideal transformer (it is a good approximation in this case) we have
2 2 1 1
N I N I or
( )
2 1 1 1 L
N I I N I or
1 2
1
2
L
N N
I I
N
+
, therefore 1
L
H
I
a
I
+ . If we now
compare the power rating of the two transformers we have:
( )
1 2 1 2
2 1 1 1
1
1 1
auto
w
V V I S N
S V I N a

+
+ + . If 1 a < then the power rating to the autotransformer
is higher than the conventional two winding transformer. This is known as the power
rating advantage of the autotransformer. In addition to this, the autotransformer has the
advantages of higher efficiency, lower internal impedance and smaller size (for the same
rating). In three phase systems this transformer is used frequently provided the value of
the voltage ratio from one side to the other does not exceed three (i.e. 3 a ).

Example 3.5

A two-winding transformer is rated 60 kVA, 240/1200 V, 60 Hz. When operated as a
conventional two winding transformer at rated load, 0.8 power factor, its efficiency is
0.96. This transformer is to be used as a 1440/1200 V step-down autotransformer.
(a) Assuming ideal transformer find the transformer kVA rating when used as an
autotransformer.
(b) Find the effiecnecy with the kVA loading of part (a) and 0.8 power factor.

The Matlab program follows:

S_2w = 60;
V1 = 240; V2 = 1200;
I1= S_2w*10^3/V1; I2 = S_2w*10^3/V2;
IL = I1 + I2;
S_auto = V2*IL*10^-3
Ploss = S_2w*0.8*(1 - 0.96)/0.96
Eff = S_auto*.8/(S_auto*0.8 + Ploss)

S_auto =
360
Ploss =
2.0000
Eff =
0.9931


We shall skip section 3.12.

Below is a short program to plot the three phase voltages of amplitude one at 60 Hz.

f=60;
w=2*pi*f;
t=0:0.0001:0.02;

9
T=1000*t;
a=cos(w*t);
b=cos(w*t+2*pi/3);
c=cos(w*t-2*pi/3);
plot(T,a,T,b,T,c),grid, xlabel('milliseconds'),
ylabel('Amplitude'), title('a is blue, b is red, c is green')

0 2 4 6 8 10 12 14 16 18 20
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
milliseconds
A
m
p
l
i
t
u
d
e
a is blue, b is red, c is green


Chapter 3. Generator and Transformer Models;
The Per-Unit System (Part III)


3.13 The Per-Unit System

Let us say that the measure unit of age is 80 years (this is the base unit). Then a person
who is 40 years old is half of the base unit. So we say the person's age is 0.5 per unit.

age in years
Age per unit
age base in years
=
Thus the age per unit is dimensionless. In power systems there are four base quantities
required to define a per unit system. These are: power, voltage, current and impedance.
Given two of these, all others can be defined using these two. In general for any
quantity:

actual quantity
Quantity per unit
base value of quantity
=
And for power systems we have:

pu pu pu pu
B B B B
S V I Z
S V I Z
S V I Z
= = = =
N.B. The value of the numerators above may be a complex number, however, the
denominator (the base value) is a positive real number.

Normally we select a three-phase power base (
B
S or
B
MVA ) and a line-to-line voltage
base (
B
V or
B
kV ). From these two the other bases can be computed using circuit laws
thus:

3
3
3
B B
B
B
B
S S
I
V
V
= =
and

3
B
B
B
V
Z
I
=
Other forms for
B
Z follow from the above equations:

( )
( )
2
2
B
B
B
B
B
B
V
Z
S
kV
Z
MVA
=
=


Note that
B
V is line to line voltage base! and
B
I is line current base. Also,
B
S is 3-phase
power base. Thus in dealing with per-phase quantities, the 3 and 3 need to be taken
into account.

2
Per unit quantities obey the circuit laws, thus:

*
pu pu pu
S V I =
and

pu pu pu
V Z I =
In terms of phase quantities, the three phase complex power to a load is given by:

*
( 3 )
3
L P P
S V I

=
and the phase current is related to the phase voltage by the load impedance thus:

P
P
P
V
I
Z
=
Using this value of phase current in the equation above we have (solving for impedance):

( ) ( )
2 2
* *
3 3
3
P L L
P
L L
V V
Z
S S

= =
Therefore the phase impedance in per-unit is given by:

( )
2
*
3
p L L
pu
B B L
Z V
Z
Z S Z

= =
and since
2
B
B
B
V
Z
S
= (see last page) the above expression becomes:

( ) ( )
2 2
* *
3 3
p L L
L L B
pu
B B B L L
Z V
V S
Z
Z S Z V S


= = =
or

( )
2
*
pu
pu
L pu
V
Z
S
=



3.14 Change of Base

Usually if none are specified, the pu values given are on nameplate ratings as base. For
example a generator whose impedance is 0.4 pu and whose ratings are 110 MVA and 24
kV, then the base MVA is assumed 110 MVA and base line to line voltage is 24 kV.
Often the base for the system is different from the base for each particualr generator or
transformer, hence it is important to be able to express the pu value in terms of different
bases. This is derived below.

Let
old
pu
Z be the per-unit impedance on the power base
old
B
S and voltage base
old
B
V . We
would like to find
new
pu
Z on the new power base
new
B
S and voltage base
new
B
V . Let Z

be
the "ohmic" value of the impedance. Then we have:
3

( )
2
old
old B
pu old
old
B
B
Z S
Z Z
Z
V

= =
and similarly

( )
2
new
new B
pu new
new
B
B
Z S
Z Z
Z
V

= =
By comparison of these two expressions it is evident that:

2
new old
new old
B B
pu pu
old new
B B
S V
Z Z
S V

=



This last expression is very useful in transforming per-unit values from one set of bases
to another.

Example. A transformer has a voltage rating of 110/220 V and a power rating of 1000
VA. The impedance of the transformer on the low voltage side is 5 33 j + . Find the per-
unit impedance viewed from each side assuming the bases are equal to the voltages and
power ratings on each side.

Solution using Matlab:

From the give quantities:

VB1=110; VB2=220; SB1=1000; SB2=1000;
Z1=5+j*33; Z2=Z1*4;
ZB1=VB1^2/SB1; ZB2=VB2^2/SB2;
Zpu1=Z1/ZB1
Zpu2=Z2/ZB2

Zpu1 =
0.4132 + 2.7273i
Zpu2 =
0.4132 + 2.7273i

Thus, with proper selection of
base
S and
base
V on each side of a transformer, the per-unit
impedance of the transformer becomes the same viewed from either side! Thus the
transformer may be removed and replaced with its per-unit impedance (but we have to
keep in mind that now there are two different values of voltage base, one on each side of
the transformer).

Rules for selection of voltage bases around a transformer: Let the turns ratio of the
transformer be
1
2
N
N
= , and suppose there is a connection gain of k . If this is a Y Y
or connected transformer, then 1 k = . If it is a Y connection, then 3 k = . If
it is Y then 1/ 3 k = . Select a voltage base on side 1, say
1 B
V . Then the voltage
base on side 2 must be:
2 1 B B
V k V = . The power base, of course is the same everywhere
in the power system.
4

Example 3.7

The one line diagram of a power system is shown in fig. 3.29 page 92. Select a common
base of 100 MVA and 22 kV on the generator side. Draw an impedance diagram with all
impedances including the load impedance marked in per unit. The manufacturer's data
for each device is given on page 92. The three phase load on bus 4 absorbs 57 MVA, 0.6
power factor lagging at 10.45 kV. Line 1 and line 2 have reactances of 48.4 and
65.43, respectively.

Outline of solution: first find the voltage base in all sections of the system. In this case
there are five sections. Note that in this case there is no mention of how the transformers
are connected (i.e. Y or delta). Hence we consider only the turns ratio (otherwise there
would be some 3 factors as well).

The Matlab program follows:

VLL = 10.45;
VB1 = 22; VB2 = 22*220/22; VB4 = 220*11/220; VB5 = 22*110/22;
Xg = 0.18*100/90
Xt1 = 0.1*100/50
Xt2 = 0.06*100/40
Xt3 = 0.064*100/40
Xt4 = 0.08*100/40
Xm = 0.185*(100/66.5)*(10.45/11)^2
ZB2 = (220)^2/100
ZB5 = (110)^2/100
Xline1 = 48.4/484
Xline2 = 65.43/121
SL = 57*(0.6 + j*sin(acos(0.6)))
ZL = VLL^2/conj(SL)
ZB4 = (11)^2/100
ZLpu = ZL/ZB4

Xg =
0.2000
Xt1 =
0.2000
Xt2 =
0.1500
Xt3 =
0.1600
Xt4 =
0.2000
Xm =
0.2511
ZB2 =
484
ZB5 =
121
Xline1 =
0.1000
Xline2 =
5
0.5407
SL =
34.2000 +45.6000i
ZL =
1.1495 + 1.5327i
ZB4 =
1.2100
ZLpu =
0.9500 + 1.2667i

Note these are then incorporated into the impedance diagram on page 94. Note there are
no transformers showing! Just an elementary one phase diagram with impedances. To
go back to actual values, we need to know the MVA
B
and the V
B
at every point in the
system. As stated earlier, the MVA
B
is the same everywhere, but the V
B
changes as we
go from one side of a transformer to the other.



Example 3.8

The motor of example 3.7 operates at full load 0.8 power factor leading at a terminal
voltage of 10.47 kV.
(a) Determine the voltage at the generator bus.
(b) Determine the generator and motor emfs.

The solution is found using ordinary circuit analysis. Of course, the answers will be "pu"
all the way!

The Matlab program follows:

ZL = 0.95 + j*1.2667;
V4=10.45/11
Sm = 66.5/100*(0.8 - j*.6)
Im = conj(Sm)/conj(V4)
IL = V4/ZL
I = Im + IL
X11 = 0.45*0.9/(0.45+0.9);
V1 = V4+j*X11*I; V1M=abs(V1), V1ang=angle(V1)*180/pi
Eg = V1 + j*0.2*I; EgM=abs(Eg), Egang=angle(Eg)*180/pi
Em = V4 - j*0.25*Im; EmM=abs(Em), Emang=angle(Em)*180/pi

V4 =
0.9500
Sm =
0.5320 - 0.3990i
Im =
0.5600 + 0.4200i
IL =
0.3600 - 0.4800i
I =
0.9200 - 0.0600i
V1M =
6
1.0066
V1ang =
15.9139
EgM =
1.0826
Egang =
25.1445
EmM =
1.0642
Emang =
-7.5591


Chapter 5. Line Model and Performance


5.1 Introduction

We shall represent lines as a "pi" section with
lumped parameters. There are three cases: short,
medium and long lines.

In the short line model (under 50 miles), we
assume the shunt capacitance (the legs of the )
are so small that they are open circuit (i.e. neglected). This leaves only the series RL
branch.

In the medium line model (50 to 250 miles), we assume that the capacitance may be
represented as two capacitors (legs of the ) each equal to half the line capacitance. This
is known as the nominal model.

In the long line model (over 250 miles), we assume the line has distributed parameters
instead of lumped parameters. This yields exact results. After finding the exact model,
an equivalent is found. This is called the equivalent model because it has lumped
parameters which are adjusted so that they are equivalent to the exact distributed
parameters model.

5.2 Short Line Model

A short line with a load is shown (per phase) in the
figure to the right. Using circuit theory we
have:
*
*
R
R
R
S
I
V
. We also know
S R R
V V ZI + . Note that
S R
I I . Now we will find the transmission parameters
(or ABCD parameters) of the short line:

For a 2-port shown as "ABCD" the equations are as
follows:
S R
S R
V V A B
I I C D
1 1 1

1 1 1
] ] ]
. Comparing with the
equations for the short line above, we see that: 1 A ,
B Z , 0 C and 1 D . This completes the equations
needed to represent the short line. The voltage regulation in percent is given by:

( ) ( )
( )
100
R NL R FL
R FL
V V
Percent VR
V

Z = R+jX
V
S
+
_
V
R
+
_
S
R
I
S
I
R
ABCD
V
S
+
_
V
R
+
_
I
S
I
R
V
S
+
_
V
R
+
_
I
S
I
R
Z = R+jX
Y/2
Y/2
I
L
2
At no-load 0
R
I and thus
( ) S R NL
V AV , and since 1 A ,
( ) R NL S
V V . This conclusion is
easily made by observation of the circuit above.

Example 5.1

A 220-kV, three phase transmission line is 40 km long. The resistance per phase is
0.15 per km and the inductance per phase is 1.3263 mH. The shunt capacitance is
negligible. Use the short line model to find the voltage and power at the sending end and
the voltage regulation and efficiency when the line is supplying three phase load of

(a) 381 MVA at 0.8 power factor lagging at 220 kV.
(b) 381 MVA at 0.8 power factor leading at 220 kV.

The Matlab program follows:

VRLL=220; VR = VRLL/sqrt(3);
Z = (0.15+j*2*pi*60*1.3263e-3)*40;
disp('(a)')
SR=304.8+j*228.6;
IR = conj(SR)/(3*conj(VR)); IS = IR;
VS = VR + Z*IR;
VSLL = sqrt(3)*abs(VS)
SS = 3*VS*conj(IS)
REG = (VSLL - VRLL)/VRLL*100
Eff = real(SR)/real(SS)*100
disp('(b)')
SR=304.8-j*228.6;
IR = conj(SR)/(3*conj(VR)); IS = IR;
VS = VR + Z*IR;
VSLL = sqrt(3)*abs(VS)
SS = 3*VS*conj(IS)
REG = (VSLL - VRLL)/VRLL*100
Eff = real(SR)/real(SS)*100

(a)
VSLL =
250.0186
SS =
3.2280e+002 +2.8858e+002i
REG =
13.6448
Eff =
94.4252
(b)
VSLL =
210.2884
SS =
3.2280e+002 -1.6862e+002i
REG =
-4.4144
Eff =
94.4252

3
5.3 Medium Line Model

As the length of the line is increased to over 50 miles (but less than 150 miles) the
medium line model is used. This is also known as the nominal model as shown below.
The admittance Y and impedance Z shown are shown for the whole line, per-phase (and
not per mile). For the "pi" model shown we can
write the following equations:
2
L R R
Y
I I V + and
S R L
V V ZI + . Using the first equation in the
second we have: 1
2
S R R
ZY
V V ZI
_
+ +

,
. We also
see that
2
S L S
Y
I I V + and thus we have: 1
2 2 2
1 1
4 2
S R R R R
R R
Y Y ZY
I I V V ZI
ZY ZY
Y V I
1
_
+ + + +
1
, ]
_ _
+ + +

, ,
.
Thus we now have the ABCD parameters:

1
2
1 1
4 2
ZY
A B Z
ZY ZY
C Y D
_
+

,
_ _
+ +

, ,

In general for a symmetrical 2-port A D and 1 AD BC (the determinant of the
ABCD matrix in this case is unity). Thus it is easy to find the inverse relationship:

S R
S R
V V D B
I I C A
1 1 1

1 1 1

] ] ]

Note the two Matlab functions for computing the ABCD and ZY parameters:

[ ] ( )
, , 2 , , , , , Z Y ABCD rlc abcd r L C g f Length

[ ] ( )
, , 2 , , Z Y ABCD zy abcd z y Length
Note also that f is frequency in Hz and g is conductance to ground per unit length (zero
in the specific case of the medium line model). Note also that , r g, z, and y are all per
unit length of line. The series impedance is
( )
Z r j L R jX + + l and the shunt
admittance is
( )
Y g j C + l .

Example 5.2

A 345-kV three-phase line is 130 km long. The resistance per phase is 0.036 per km
and the inductance per phase is 0.8 mH per km. The shunt capacitance is 0.0112 F per
km. The receiving end load is 270 MVA with 0.8 power factor lagging at 325 kV. Use
the medium line model to find the voltage and power at the sending end and the voltage
regulation.

V
S
+
_
V
R
+
_
I
S
I
R
Z = R+jX
Y/2
Y/2
I
L
4
Note that we cannot run this program from within MS-Word since it has interactive
statements. A listing of the program for example 5.2 is shown below. Run example 5.2
in the Matlab window (chp5ex2). DO NOT EXECUTE IN THIS NOTEBOOK!
(Reason: edit the function "rlc2abcd.m", it has interactive code!)

r = .036; g = 0; f = 60;
L = 0.8; % milli-Henry
C = 0.0112; % micro-Farad
Length = 130; VR3ph = 325;
VR = VR3ph/sqrt(3) + j*0; % kV (receiving end phase voltage)
[Z, Y, ABCD] = rlc2abcd(r, L, C, g, f, Length);
AR = acos(0.8);
SR = 270*(cos(AR) + j*sin(AR)); % MVA (receiving end power)
IR = conj(SR)/(3*conj(VR)); % kA (receiving end current)
VsIs = ABCD* [VR; IR]; % column vector [Vs; Is]
Vs = VsIs(1);
Vs3ph = sqrt(3)*abs(Vs); % kV(sending end L-L voltage)
Is = VsIs(2); Ism = 1000*abs(Is); % A (sending end current)
pfs= cos(angle(Vs)- angle(Is)); % (sending end power factor)
Ss = 3*Vs*conj(Is); % MVA (sending end power)
REG = (Vs3ph/abs(ABCD(1,1)) - VR3ph)/VR3ph *100;
fprintf(' Is = %g A', Ism), fprintf(' pf = %g\n', pfs)
fprintf(' Vs = %g L-L kV\n', Vs3ph)
fprintf(' Ps = %g MW', real(Ss)),
fprintf(' Qs = %g Mvar\n', imag(Ss))
fprintf(' Percent voltage Reg. = %g\n', REG)

Example 5.3

A 345-kV, three-phase transmission line is 130 km long. The series impedance is
0.036 0.3 z j + per phase per km. The shunt admittance is
6
4.22 10 y j

siemens
per phase per km. The sending end voltage is 345-kV and the sending end current is 400-
A at 0.95 power factor lagging. Use the medium line model to find the voltage, current
and power at the receiving end and the voltage regulation.

Note that we cannot run this program from within MS-Word since it has interactive
statements. A listing of the program for example 5.3 is shown below. Run example 5.3
in the Matlab window (chp5ex3). DO NOT EXECUTE IN THIS NOTEBOOK! The
culprit is the command "zy2abcd" which has interactive code.

z = .036 + j* 0.3; y = j*4.22/1000000; Length = 130;
Vs3ph = 345; Ism = 0.4; %KA;
As = -acos(0.95);
Vs = Vs3ph/sqrt(3) + j*0; % kV (sending end phase voltage)
Is = Ism*(cos(As) + j*sin(As));
[Z,Y, ABCD] = zy2abcd(z, y, Length);

VrIr = inv(ABCD)* [Vs; Is]; % column vector [Vr; Ir]
Vr = VrIr(1);
Vr3ph = sqrt(3)*abs(Vr); % kV(receiving end L-L voltage)
Ir = VrIr(2); Irm = 1000*abs(Ir);% A (receiving end current)
pfr= cos(angle(Vr)- angle(Ir)); % (receiving end power factor)
5
Sr = 3*Vr*conj(Ir); % MVA (receiving end power)
REG = (Vs3ph/abs(ABCD(1,1)) - Vr3ph)/Vr3ph *100;
fprintf(' Ir = %g A', Irm), fprintf(' pf = %g\n', pfr)
fprintf(' Vr = %g L-L kV\n', Vr3ph)
fprintf(' Pr = %g MW', real(Sr))
fprintf(' Qr = %g Mvar\n', imag(Sr))
fprintf(' Percent voltage Reg. = %g\n', REG)


5.4 Long Line Model

This is a distributed parameters model of the line. It is very exact and gives the voltage
and current anywhere along the line. The distance x is measured from the receiving end
towards the sending end. If we let the x l (the length of the line) the equations for the
long line model have the following ABCD parameters:

cosh sinh
1
sinh cosh
c
c
A B Z
C D
Z




l l
l l

where
c
Z is the line characteristic impedance given by
c
z
Z
y
and j zy + .
It is now possible to find an accurate model whose ABCD parameters are equal in
value to the ones shown above. Without proof the results are outlined below:

sinh
sinh
1 tanh / 2
tanh
2 2 / 2
c
c
Z Z Z
Y Y
Z


l
l
l
l
l
l

where the primed quantities are for the exact or equivalent model, and the unprimed
quantities are those for the nominal . This way we can transform the nominal to an
exact or equivalent . The Matlab functions mentioned earlier (rlc2abcd and zy2abcd)
with option 2 will perform the equivalent model for a long line.

Example 5.4

A 500-kV, three phase transmission line is 250 km long. The series impedance is
0.045 0.4 z j + per phase per km and the shunt admittance is
6
4 10 y j

siemens
per phase per km. Evaluate the equivalent model and the transmission matrix.

Below is the program for example 5.4:

z = 0.045 + j*.4; y = j*4.0/1000000; Length = 250;
gamma = sqrt(z*y); Zc = sqrt(z/y);
A = cosh(gamma*Length); B = Zc*sinh(gamma*Length);
C = 1/Zc * sinh(gamma*Length); D = A;
ABCD = [A B; C D]
Z = Zc * sinh(gamma*Length)
Y = 2/Zc * tanh(gamma*Length/2)
6

ABCD =
0.9504 + 0.0055i 10.8778 +98.3624i
-0.0000 + 0.0010i 0.9504 + 0.0055i
Z =
10.8778 +98.3624i
Y =
0.0000 + 0.0010i

Chapter 6. Power Flow Analysis


6.1 Introduction

The system is assumed 3-phase balanced operating in a steady state and stable condition.
It is represented by a single phase diagram on a per-unit basis, with a system wide MVA
base, and a voltage base properly chosen on each side of every transformer. The base
MVA and base V are specified/known everywhere in the system.

The most common way to represent such a system is to use the node-voltage method.
Given the voltages of generators at all generator nodes, and knowing all impedances of
machines and loads, one can solve for all the currents in the typical node voltage analysis
methods using Kirchoff's current law. First the generators are replaced by equivalent
current sources and the node equations are written in the form:
I YV =
where I is the injected current vector, Y is the admittance matrix and V is the node
voltage vector. These equations are easy to write by inspection of the circuit.

The problem is not so simple in real power circuits and systems. Usually in a power
system the complex power may be known at load nodes, and sometimes on generator
nodes only the real power and voltage are known. Thus not enough variables are known
to solve an equation of the form I YV = . In fact, since the power is a nonlinear function
of the current and voltage, the solution of the resulting equations (while it may exist) is
not easy! In fact there is no known analytical method to find the solution. As a result
iterative techniques are used to find the solution (voltages, currents, etc.). The nonlinear
set of equations which are generated are called power flow equations. The solution of
such equations results in a power flow study or load flow analysis. Such studies are the
backbone of power system studies, for analysis, design, control, and economic operation
of the power system. They are also essential for transient analysis of the system.


6.2 Bus Admittance Matrix

1. The first step is to number all the nodes of the system from 0 to n . Node 0 is the
datum or reference node (or ground node).
2. replace all generators by equivalent current sources in parallel with an admittance.
3. Replace all lines, transformers, loads to equivalent admittances whenever possible.
Knowing a load in MVA but not knowing the operating voltage, makes it impossible
to change the load to an admittance. The rule for this is simple: 1/ y z = where
and y z are generally complex numbers.
4. The bus admittance matrix Y is then formed by inspection as follows (this is similar
to what we learned in circuit theory):
sum of admittances connected to node
ii
y i =
and
sum of admittances connected from node to node
ij ji
y y i j = =

5. The current vector is next found from the sources connected to nodes 0 to n . If no
sourde is connected, the injected current would be 0.
6. The equations which result are called the node-voltage equations and are given the
"bus" subscript in power studies thus:

bus bus bus
I Y V =
The inverse of these equations results in the set
1
bus bus bus
V Y I

= . It is emphasized that the


matrix
1
bus
Y

is not the same as the matrix Z which results from solving a circuit using
mesh equations. To clearly show this difference we define
1
bus bus
Z Y

= . It is noted that the


matrix
bus
Y is the same as the Y matrix obtained from circuit theory. Thus
bus
Y Y =
however
bus
Z Z .

It is also noted that in general, solving a large set of linear equations is never done using
the matrix inverse. Finding the inverse involves more work than is needed to find a
solution and the resulting solutions are less accurate. Usually Gauss elimination or a
similar factorization scheme is used in a direct solution. Iterative solutions are also very
effective in larger systems.

It is further noted that in a real power system with hundreds of nodes, each node is rarely
connected to more than two or three other nodes, thus most of the elements of the
admittance matrix are zero (the admittance matrix is sparse). Special techniques exist to
solve systems with sparse matrices.

In a power system the
bus
Z is built up directly from the power system (see section 9.5)
and matrix inversion is not used. Also, we have a routine to form the admittance matrix
called ybus used as follows: Y = ybus(zdata) where "zdata" is a matrix of entries
having n-rows and 4 columns. Column 1 and 2 indicate the nodes between which the
impedance exists (nonzero), and columns 3 and 4 give the real and imaginary parts of the
impedance. The result is the bus admittance matrix
bus
Y .

Example 6.1. The emfs shown in figure 6.1 page 190 are
1
1.1 0 E =
o
and
2
1.0 0 E =
o
.
Use the function Y = ybus(zdata) to obtain the bus admittance matrix. Find the
bus impedance matrix by inversion and solve for the bus voltages.

Tha Matlab program is shown below:

% From To R X
z = [ 0 1 0 1.0
0 2 0 0.8
1 2 0 0.4
1 3 0 0.2
2 3 0 0.2
3 4 0 0.08];
[Ybus] = ybus(z) % bus admittance matrix
Ibus = [-j*1.1; -j*1.25; 0; 0]; % vector of injected bus
currents
Zbus = inv(Ybus) % bus impedance matrix
Vbus = Zbus*Ibus
%Vbus = Ybus\Ibus

Note: the last statement which is commented out is the better way of doing it! Why?

Do not execute chp6ex1 from this workbook. Instead, go to the Matlab window and give
the command to get the output.


6.3 Solution of Nonlinear Algebraic Equations

The most common methods for solving nonlinear algebraic equations are Gauss-Seidel,
Newtow-Rahpson, and quasi-Newton-Raphson methods. We start with one dimensional
equations and then generalize to n-dimensional equations.


6.3.1 Gauss-Seidel Method

This method is also known as the method of successive displacements. Consider the
nonlinear equation ( ) 0 f x = . The equation is broken into two parts thus: ( ) x g x = . We
assume
( ) 0
x is an initial "guess" of the solution, then "refine" the solution using:

( ) ( )
( )
1 0
x g x =

This process is repeated thus

( ) ( )
( )
2 1
x g x =
and on the
th
n iteration we have:
( ) ( )
( )
1 n n
x g x

= . If this process is convergent, then the
successive solutions approach a value which is declared as the solution. Thus if at some
step 1 k + we have
( ) ( ) 1 k k
x x
+
where is the desired "accuracy", then we claim the
solution has been found to the accuracy specified.

Example 6.2 Use the Gauss-Seidel method to obtain the roots of the equation

( )
3 2
6 9 4 0 f x x x x = + =
First the equation is expressed in a different form thus

( ) ( )
3 2
1
6 4
9
x x x g x = =
And the iteration can proceed. Take a good look at the shape of the iterations! Below
is the program showing the process graphically (later showing how to do it iteratively).
clear; x=0:.01:4.5;
g=-1/9*x.^3+6/9*x.^2+4/9;
k=1;
dz=10;
z=2;
r(k)=z;
s(k)=z;
while dz >.001
k=k+2;
r(k-1)=z;
p=-1/9*z^3+6/9*z^2+4/9;
s(k-1)=p;
dz=abs(z-p);
z=p;
r(k)=z; s(k)=z;
end
plot(x,g,'-',x,x,'-', r,s,'-'),grid
xlabel('x')
text(0.8, 4.2,'g(x) =-1/9x^3+6/9x^2+4/9')
text(3.2, 3.0,'x')

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
x
g(x) =-1/9x
3
+6/9x
2
+4/9
x


Note: try running the program above with an initial "guess" of z = 0 and z = 0.8
Does this process converge? always? to a unique solution?

Next, plot the function and find the roots using Matlab:

roots([1; -6; 9; -4])
x=0:0.01:4.5;
zz=zeros(size(x)); % plot the x-axis
f=x.^3-6*x.^2+9.*x-4;
plot(x,f,x,zz),grid

ans =
4.0000
1.0000 + 0.0000i
1.0000 - 0.0000i
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
-4
-2
0
2
4
6
8


Note the double root at x = 1.

Now the program to do this iteratively:

dx=1; % Change in variable is set to a high value
x=2; % Initial estimate
iter = 0; % Iteration counter
disp('Iter g dx x') % Heading for results
while abs(dx) >= 0.001 & iter < 100 % Test for convergence
iter = iter + 1; % No. of iterations
g = -1/9*x^3+6/9*x^2+4/9 ;
dx = g-x; % Change in variable
x = x + dx; % Successive approximation
fprintf('%g', iter), disp([g, dx, x])
end

Iter g dx x
1 2.2222 0.2222 2.2222
2 2.5173 0.2951 2.5173
3 2.8966 0.3793 2.8966
4 3.3376 0.4410 3.3376
5 3.7398 0.4022 3.7398
6 3.9568 0.2170 3.9568
7 3.9988 0.0420 3.9988
8 4.0000 0.0012 4.0000
9 4.0000 0.0000 4.0000


Remarks: The roots found depend on the starting point! There may be other roots
Moreover, there is no guarantee the method is converging unless one checks on the
convergence by means of some algorithm.

The Gauss-Seidel method can be expressed with a parameter known as an acceleration
factor, thus:

( ) ( ) ( )
( )
( ) 1 k k k k
x x g x x
+

= +


Note that if 1 = the method becomes the non-accelerated method mentioned earlier.

Example 6.3. Find the roots of the equation used earlier but with an acceleration factor of
1.25.

First a graphical solution using Matlab:

clear; x=0:.05:4.5;
alpha=1.25;
g=-1/9*x.^3+6/9*x.^2+4/9;
k=1;
dz=10;
z=2;
r(k)=z;
s(k)=z;
while dz >.001
k=k+2;
r(k-1)=z;
p=-1/9*z^3+6/9*z^2+4/9;
s(k-1)=p;
dz=abs(z-p);
z=z+alpha*(p-z);
r(k)=z; s(k)=p;
end
plot(x,g,'-',x,x,'-', r,s,'-'),grid
xlabel('x')
text(0.7, 4.1,'g(x) = -1/9x^3+6/9x^2+4/9')
text(3.2, 3.0,'x')

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
x
g(x) = -1/9x
3
+6/9x
2
+4/9
x


Note how at each step we choose more than "x" trying to go "further" in the right
direction. This can also produce an unstable result, especially if 1 ? .

Next the numerical solution using Matlab:

%chp6fig4 % Graphical display for Example 6.3
clear; dx=1; % Change in variable is set to a high value
x=2; % Initial estimate
iter = 0; alpha=1.25; % Iteration counter, alpha
disp('Iter g dx x') % Heading for results
while abs(dx) >= 0.001 & iter < 100 % Test for convergence
iter = iter + 1; % No. of iterations
g = -1/9*x^3+6/9*x^2+4/9;
dx = g-x; % Change in variable
x = x +alpha*dx; % Successive approximation with alpha accel. factor
fprintf('%g', iter), disp([g, dx, x])
end

Iter g dx x
1 2.2222 0.2222 2.2778
2 2.5902 0.3124 2.6683
3 3.0801 0.4118 3.1831
4 3.6157 0.4326 3.7238
5 3.9515 0.2277 4.0084
6 4.0000 -0.0085 3.9978
7 4.0000 0.0022 4.0005
8 4.0000 -0.0005 3.9999

Next the process is generalized for n equations in n variables:

( )
( )
( )
1 1 2 1
2 1 2 2
1 2
, , ,
, , ,
, , ,
n
n
n n n
f x x x c
f x x x c
f x x x c
=
=
=
L
L
LLLLLLLL
L

Solving for one variable from each equation, the system of equations car be represented
as follows:

( )
( )
( )
1 1 1 1 2
2 2 2 1 2
1 2
, , ,
, , ,
, , ,
n
n
n n n n
x c g x x x
x c g x x x
x c g x x x
= +
= +
= +
L
L
LLLLLLLLLL
L

First a starting solution is assumed to be
( ) ( ) ( )
( )
0 0 0
1 2
, , ,
n
x x x L . Now there are two ways to
proceed:

1. Gauss iteration: use the starting estimate
( ) ( ) ( )
( )
0 0 0
1 2
, , ,
n
x x x L in the equations to
compute a first iterate
( ) ( ) ( )
( )
1 1 1
1 2
, , ,
n
x x x L then repeat the process using the first iterate
to get a second iterate
( ) ( ) ( )
( )
2 2 2
1 2
, , ,
n
x x x L . This is repeated till the solutions converge.

2. Gauss-Seidel iteration: use the starting estimate
( ) ( ) ( )
( )
0 0 0
1 2
, , ,
n
x x x L in the first equation
ONLY to get a new estimate for
( ) 1
1
x . Now use
( ) ( ) ( )
( )
1 0 0
1 2
, , ,
n
x x x L in the second
equation to get an estimate for
( ) 1
2
x . Now use
( ) ( ) ( )
( )
1 1 0
1 2
, , ,
n
x x x L in the third equation
to get an estimate for
( ) 1
3
x and so on till the whole set
( ) ( ) ( )
( )
1 1 1
1 2
, , ,
n
x x x L is found. The
process is repeated till the solutions converge.

An acceleration factor can be used as before, thus at each step upgrade the estimate as
follows

( ) ( ) ( ) ( )
( )
1 1 k k k k
i i ical i
x x x x
+ +
= +
Clearly if 1 = then the iteration is the same as before. Usually 1 but not much
larger than unity.
Chapter 6. Power Flow Analysis (Part II)


6.3.2 Newton-Raphson Iteration

In this method both the function
( )
f x and its derivative
( )
f x are needed to perform
the iteration. Graphically, this is shown below for
( )
3 2
6 9 4 f x x x x + . In this
example, the derivative is easily found as
( )
2
3 12 9 f x x x + . Take a look at the
program chp6fig5 using the Matlab Editor.

chp6fig5

0 1 2 3 4 5 6
-10
0
10
20
30
40
50
x
f(x) = x
3
-6x
2
+9x-4


Using the slope instead of the function to get the new estimate as shown above. This is a
linear estimate of the root of
( )
0 f x . The equation used for the iterations is given by:

( ) ( )
( )
( )
( )
( )
1
k
k k
k
f x
x x
f x
+


where a starting solution
( ) 0
x is needed to start the iteration. Note that there is no
guarantee of convergence, and, if the iteration converges, there may be other roots as
well.

In general, let
( )
f x c . Expanding
( )
f x in a Taylor series about the initial estimate
( ) 0
x results in

( )
( )
( )
( )
( )
( )
( )
0
0
2
2
0 0 0
2
1
HOT
2!
df d f
f x x x c
dx dx
_
_
+ + + +

,
,
L
keeping only the linear terms and neglecting the higher order terms we have

( )
( )
( )
( )
0
0 0
df
f x x c
dx
_
+

,
;
which can be written as
( )
( )
( )
( )
0
0 0
df
c f x x
dx
_


,
; . If we define
( ) ( )
( )
0 0
c c f x @ then
( )
( )
( )
0
0 0
df
c x
dx
_


,
; . Thus we have
( )
( )
( )
0
0
0
c
x
df
dx


_

,
and a new estimate for
( ) 1
x is now
given by:
( ) ( ) ( ) ( )
( )
( )
0
1 0 0 0
0
c
x x x x
df
dx

+ +
_

,
. This process can be repeated several times,
at the (k+1)
th
iteration we have:

( ) ( )
( )
k k
c c f x @

( )
( )
( )
k
k
k
c
x
df
dx


_

,


( ) ( ) ( ) 1 k k k
x x x
+
+
The middle equation above can be written as:

( )
( )
( )
k
k k
df
c x
dx
_


,

We define the derivative at step k as follows:
( )
( ) k
k
df
j
dx
_

,
@ thus we have:

( ) ( ) ( ) k k k
c j x
In summary: given the nonlinear equation
( )
f x c or
( )
0 f x c , the curve is
approximated at
( ) k
x by a linear approximation which is the tangent to the curve. The
intersection of the linear approximation with the x-axis is the next estimate
( ) 1 k
x
+
.

Example 6.4. Use the N-R method to find the root of the equation
3 2
( ) 6 9 4 f x x x x + assume an initial value of
( ) 0
6 x .

This is demonstrated graphically in the plot shown above. It is next solved by iteration.

dx=1; % Change in variable is set to a high value
x=6; % Initial estimate
iter = 0; % Iteration counter
disp('iter Dc J dx x') % Heading for result
while abs(dx) >= 0.001 & iter < 100 % Test for convergence
iter = iter + 1; % No. of iterations
Dc=4 - (x^3-6*x^2+9*x); % Residual
J = 3*x^2-12*x+9; % Derivative
dx= Dc/J; %Change in variable
x=x+dx; % Successive solution
fprintf('%g', iter), disp([Dc, J, dx, x])
end

iter Dc J dx x
1 -50.0000 45.0000 -1.1111 4.8889
2 -13.4431 22.0370 -0.6100 4.2789
3 -2.9981 12.5797 -0.2383 4.0405
4 -0.3748 9.4914 -0.0395 4.0011
5 -0.0095 9.0126 -0.0011 4.0000
6 -0.0000 9.0000 -0.0000 4.0000

Note that it took only 6 iterations. This is true in almost all cases, even if the function is
very complicated and of a high order. If the method is going to converge, it will
converge within a few iterations (usually of the order of 10 or less). Note in the program
a limit of 100 iterations is stipulated.

Next the method is generalized to the n-dimensional case. The equations are:

( )
( )
( )
1 1 2 1
2 1 2 2
1 2
, , ,
, , ,
, , ,
n
n
n n n
f x x x c
f x x x c
f x x x c

L
L
LLLLLLLL
L

Expanding the functions in a Taylor series and keeping only the linear terms we have:

( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
0 0 0
0 0 0 0
1 1 1
1 1 2 1
1 2
0 0 0
0
0 0 0
2 2 2
2 1 2 2
1 2
0 0 0
0
0 0 0
1 2 1
1 2
n
n
n
n
n n n
n n
n
f f f
f x x x c
x x x
f f f
f x x x c
x x x
f f f
f x x x c
x x x
_ _ _
+ + + +


, , ,
_ _ _
+ + + +


, , ,
_ _ _
+ + + +


, , ,
L
L
L
L

which may be expressed in matrix form as follows:

( )
( )
( )
( )
( )
( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( )
( )
( )
0 0 0
1 1 1
1 2 0
0
1 1
1
0 0 0
0 0
2 2 2
2 2 2
1 2 2
0 0
0 0 0
1 2
n
n
n n
n n n
n
f f f
x x x
c f
x
f f f
x c f
x x x
x
c f
f f f
x x x
1
_ _ _
1


1
, , ,
1


1
1
_ _ _ 1
1


1


1
, , ,
1
1
1
1

1
1
]
_ _ _ 1
1

1
, , ,
]
L
L
M M
M M O M
L
1
1
1
1
1
1
]

In general, assuming we are at step k instead of 0, this may be written compactly as:

( ) ( ) ( ) k k k
C J X
which can be solved for
( ) k
X thus:

( ) ( ) ( )
1
k k k
X J C

1

]

and thus the next iterate is given by:

( ) ( ) ( ) 1 k k k
X X X
+
+
where

( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
1 1
1
2 2 2
1 1 1
1 2
2 2 2
1 2 2
1 2
and
k
k
k
k
k k
k k
n
n n
k k k
n
k k k
k
k k k
n n n
n
c f
x
x c f
X C
x
c f
f f f
x x x
f f f
J
x x x
f f f
x x x
1
1

1
1
1
1

1
1
1
1
1
1


] 1
]
_ _ _


, , ,
_ _ _


, , ,
_ _ _


, , ,
M M
L
L
M M O M
L
1
1
1
1
1
1
1
1
1
1
1
1
]

( ) k
J is the generalized derivative in n-dimensions. It is called the Jacobian matrix.
Elements of this matrix are partial derivatives of the nonlinear functions with respect to
the variables and evaluated at
( ) k
X . It is assumed that this Jacobian has an inverse at
each step of the iteration. Thus, instead of dealing with nonlinear equations, the
equations are first linearized, and solved for each iteration. Note that inversion of the
Jacobian is not needed if we only want the solution of the linear matrix equation. In
Matlab we use \ X J C in order to solve this set of equations without obtaining the
inverse of the Jacobian.


Example 6.5. Use the N-R method to find the intersections of the curves:

1
2 2
1 2
2
4
1
x
x x
e x
+
+

First a graph of the functions is shown below:

chp6fig6

Current plot held
-3 -2 -1 0 1 2 3
-3
-2
-1
0
1
2
3
x
x
2
+y
2
=4
e
x
+y=1


Next the problem is solved using the N-R iteration. The Jacobian is found from the
equations by differentiation thus:

1
1 2
2 2
1
x
x x
J
e
1

1
]

The Matlab program follows:

iter = 0; % Iteration counter
x=[0.5;-1]; % Initial estimate
Dx = [1; 1];
C=[4; 1];
disp('Iter DC Jacobian matrix Dx x');
% Heading for results
while max(abs(Dx)) >= .0001 & iter < 10 %Convergence test
iter=iter+1; % Iteration counter
f = [x(1)^2+x(2)^2; exp(x(1))+x(2)]; % Functions
DC = C - f; % Residuals
J = [2*x(1) 2*x(2) % Jacobian matrix
exp(x(1)) 1];
Dx=J\DC; % Change in variables
x=x+Dx; % Successive solutions
fprintf('%g', iter), disp([DC, J, Dx, x]) % Results
end

Iter DC Jacobian matrix Dx x
1 2.7500 1.0000 -2.0000 0.8034 1.3034
0.3513 1.6487 1.0000 -0.9733 -1.9733
2 -1.5928 2.6068 -3.9466 -0.2561 1.0473
-0.7085 3.6818 1.0000 0.2344 -1.7389
3 -0.1205 2.0946 -3.4778 -0.0422 1.0051
-0.1111 2.8499 1.0000 0.0092 -1.7296
4 -0.0019 2.0102 -3.4593 -0.0009 1.0042
-0.0025 2.7321 1.0000 0.0000 -1.7296
5 -0.0000 2.0083 -3.4593 -0.0000 1.0042
-0.0000 2.7296 1.0000 -0.0000 -1.7296

The other intersection can be found using the initial estimate [-0.5 ; 1].

Example 6.6

Starting with an initial value [1, 1, 1] x solve the following system of equations using
the N-R method.

2 2 2
1 2 3
2
1 2 2 3
1 1 3 2 3
11
3 3
6
x x x
x x x x
x x x x x
+
+
+

First we find the Jacobian, thus:

1 3
2 1 2
3 3 1 2
2 2 2
2 3
1
x x x
J x x x
x x x x

1
1
+
1
+ 1
]

The Matlab program follows:

Dx=[10; 10; 10]; % Change in variable is set to a high value
x=[1; 1; 1]; % Initial estimate
C=[11; 3; 6];
iter = 0; % Iteration counter
while max(abs(Dx))>=.0001 & iter<10; % Test for convergence
iter = iter + 1; % No. of iterations
F = [x(1)^2-x(2)^2+x(3)^2 % Functions
x(1)*x(2)+x(2)^2-3*x(3)
x(1)-x(1)*x(3)+x(2)*x(3)];
DC =C - F; % Residuals
J = [2*x(1) -2*x(2) 2*x(3) % Jacobian matrix
x(2) x(1)+2*x(2) -3
1-x(3) x(3) -x(1)+x(2)];
Dx=J\DC; %Change in variable
x=x+Dx; % Successive solution
end
iter, DC, J, Dx, x

iter =
6
DC =
1.0e-004 *
-0.3669
-0.1784
0.0957
J =
4.0000 -6.0000 8.0000
3.0000 8.0000 -3.0000
-3.0000 4.0000 1.0000
Dx =
1.0e-005 *
-0.5577
-0.1130
-0.2645
x =
2.0000
3.0000
4.0000

It is noted that the convergence of the method is quadratic, and is independent of the size
of the system being iterated. Usually convergence takes place in 10 iterations or less.





Chapter 6. Power Flow Analysis (Part III)


6.4 Power Flow Solution

Power flow studies also known as load flow are important in power system analysis.
They are used in system planning, economic scheduling, control, and future plans for the
power system. It is assumed the system is operating in a balanced steady state. The
problem is to determine the magnitude and phase angle of voltages at every node, and the
real and reactive power flow for each line.

A single phase model is used for analysis (due to balanced conditions). The per-unit
system is almost always used. Four quantities are associated with each bus. These are
voltage magnitude V , and phase angle , real power P , and reactive power Q. The
system buses are classified into three types as follows:

Slack bus One bus, known as the slack or swing bus, is taken as reference where the
magnitude and phase angle of the voltage are specified. After specifying the powers of
some other buses, this bus takes the "slack" by balancing the power equation. This bus
will be bus #1. Both P and Q need to be found for this bus.

Load buses At these buses the real and reactive powers are known. The voltage
magnitude and angle need to be found. These buses are also known as P-Q buses.

Generator buses These buses are the generator buses. At these buses the voltage
magnitude and real power are known. Other names for these buses: P-V buses, regulated
buses and voltage controlled buses. The unknown variables on these buses are the
reactive power Q and the voltage angle .


6.4.1 Power Flow Equation

Consider a typical bus in a power system as shown
in the figure on the right for bus i. Transmission
lines are represented by their equivalent models
and the impedances are converted to admittances on
a per-unit basis on a common MVA base and
properly selected kV bases. Application of KCL
we have:

( ) ( ) ( )
( )
0 1 1 2 2
0 1 1 1 2 2
i i i i i i i in i n
i i in i i i in n
I y V y V V y V V y V V
y y y V y V y V y V
+ + + +
+ + +
L
L L

y
i1
y
i2
y
in
y
i0
I
i
V
i
V
1
V
2
V
n
Bus i
2
or
0 1
n n
i i ij ij j
j j
I V y y V j i



. We also know the real and reactive power for bus i are
given by

*
i i i i
P jQ VI +
or

*
i i
i
i
P jQ
I
V


we have:

*
0 1
###
n n
i i
i ij ij j
j j
i
P jQ
V y y V j i
V




The above equation is the fundamental equation from which the power flow equations are
formulated. These will result in nonlinear algebraic equations which are solved by the
iterative techniques discussed earlier.


6.5 Gauss-Seidel Power Flow Solution

Equation ### is solved for the voltages for iteration thus:

( )
( )
( )
*
1 1,
0,
sch sch n
k
i i
ij j k
k j j i
i
i n
ij
j j i
P jQ
y V
V
V
y
+


where
sch
i
P and
sch
i
Q are the net power flows into bus i due to the current entering the
bus. For generator buses, the value of
sch
i
P and
sch
i
Q is positive when power is going
into the bus. For load buses,
sch
i
P and
sch
i
Q are positive when the power is leaving the
bus.

We can solve for the real and imaginary power using equation ###, thus:

( ) ( ) ( ) ( ) 1 *
0, 1,
Re
n n
k k k k
i i i ij ij j
j j i j j i
P V V y y V
+


1


' ;
1
]




( ) ( ) ( ) ( ) 1 *
0, 1,
Im
n n
k k k k
i i i ij ij j
j j i j j i
Q V V y y V
+


1


' ;
1
]



These three equations are the ones used for the iteration after some more manipulation of
the terms involved. Usually the elements of the matrix
bus
Y are known, hence we will
express these equations in terms of the elements of
bus
Y shown by upper case letters thus:
off diagonal elements
ij ij
Y y and diagonal elements
0,
n
ii ij
j j i
Y y


. The three
equations above become:
3

( )
( )
( )
*
1 1,
*
sch sch n
k
i i
ij j
k
k j j i
i
i
ii
P jQ
Y V
V
V
Y
+



( ) ( ) ( ) ( ) 1 *
1,
Re **
n
k k k k
i i i ii ij j
j j i
P V V Y Y V
+


1

+
' ;
1
]



( ) ( ) ( ) ( ) 1 *
1,
Im ***
n
k k k k
i i i ii ij j
j j i
Q V V Y Y V
+


1

+
' ;
1
]



It is noted that
ii
Y includes the admittance to ground.

We are interested in solving fo the voltage (magnitude and angle) at all the buses. Thus
there are 2n unknowns. However, at the slack bus we assume
1
1 0 V
o
, thus we now
have
( )
2 1 n unknowns which we need to fi nd by iteration.

For all other buses, we start the iteration with an initial estimate of 1 0 1 0 V j +
o
.

For P-Q buses we know the scheduled power components
sch
i
P and
sch
i
Q thus equation
(*) above can be used to iterate for the voltages.

For P-V buses we know
sch
i
P and
i
V , thus we can use (***) to find
sch
i
Q , then use (*) to
iterate for V . However, since
i
V is fixed, only the imaginary part of
( ) 1 k
i
V
+
is retained,
and the real part is selected so as to satisfy

( )
( )
( )
( )
2 2
2
1 1 k k
i i i
e f V
+ +
+
or

( ) ( )
( )
2
2
1 1 k k
i i i
e V f
+ +

where
( ) 1 k
i
e
+
and
( ) 1 k
i
f
+
are the real and imaginary components of the voltage, i.e.
( ) ( ) ( ) 1 1 1 k k k
i i i
V e j f
+ + +
+ .

Often an acceleration factor is employed to increase the rate of convergence, thus

( ) ( ) ( ) ( )
( )
1 k k k k
i i ical i
V V V V
+
+
where is in the range of 1.3 to 1.7.

The updated voltages are used immediately, and the whole process is repeated till the
solution converges. A good test for convergence is when

( ) ( ) 1 k k
i i
e e
+


( ) ( ) 1 k k
i i
f f
+

4
V
i
V
j
I
ij
I
ji
I
l
I
i0
I
j0
y
ij
y
i0
y
j0
where is in the range 0.00001 to 0.00005 pu.

In practical systems it is the power mismatch vector which is used to judge convergence,
thus the largest element in the column vectors and P Q is required to be smaller than
a specified value (usually of the order of 0.001 pu). Once a solution is found, the real and
reactive powers of the slack bus are computed from (**) and (***) above.


6.6 Line Flows and Losses

Having found the voltages at all buses, the next step is
to compute the line flows and losses. Consider a
typical line connecting buses i and j as shown on the
left. The total current from bus i to j is given by
( )
0 0 ij l i ij i j i i
I I I y V V y V + + . Similarly, the total
current from bus j to bus I is given by
( )
0 0 ji l j ij j i j j
I I I y V V y V + + . The complex
powers
ij
S from bus i to bus j, and
ji
S from bus j to bus i are given by

*
*
ij i ij
ji j ji
S V I
S V I


The power loss in line i j is the algebraic sum of the two power expressions above, thus

Li j ij ji
S S S +


Example 6.7

Read the example and the longhand solution in the book page 212, then return for the
Matlab solution below (chp6ex7.m):

y12=10-j*20;
y13=10-j*30;
y23=16-j*32;
V1=1.05+j*0;
iter =0;
S2=-2.566-j*1.102;
S3=-1.386-j*.452;
V2=1+j*0;
V3=1+j*0;
disp(' iter V2 V3')
for I=1:10;
iter=iter+1;
V2 = (conj(S2)/conj(V2)+y12*V1+y23*V3)/(y12+y23);
V3 = (conj(S3)/conj(V3)+y13*V1+y23*V2)/(y13+y23);
disp([iter, V2, V3])
end
V2= .98-j*.06;
5
V3= 1-j*.05;
I12=y12*(V1-V2); I21=-I12;
I13=y13*(V1-V3); I31=-I13;
I23=y23*(V2-V3); I32=-I23;
S12=V1*conj(I12); S21=V2*conj(I21);
S13=V1*conj(I13); S31=V3*conj(I31);
S23=V2*conj(I23); S32=V3*conj(I32);
I1221=[I12,I21]
I1331=[I13,I31]
I2332=[I23,I32]
S1221=[S12, S21 (S12+S13) S12+S21]
S1331=[S13, S31 (S31+S32) S13+S31]
S2332=[S23, S32 (S23+S21) S23+S32]

iter V2 V3
1.0000 0.9825 - 0.0310i 1.0011 - 0.0353i
2.0000 0.9816 - 0.0520i 1.0008 - 0.0459i
3.0000 0.9808 - 0.0578i 1.0004 - 0.0488i
4.0000 0.9803 - 0.0594i 1.0002 - 0.0497i
5.0000 0.9801 - 0.0598i 1.0001 - 0.0499i
6.0000 0.9801 - 0.0599i 1.0000 - 0.0500i
7.0000 0.9800 - 0.0600i 1.0000 - 0.0500i
8.0000 0.9800 - 0.0600i 1.0000 - 0.0500i
9.0000 0.9800 - 0.0600i 1.0000 - 0.0500i
10.0000 0.9800 - 0.0600i 1.0000 - 0.0500i
I1221 =
1.9000 - 0.8000i -1.9000 + 0.8000i
I1331 =
2.0000 - 1.0000i -2.0000 + 1.0000i
I2332 =
-0.6400 + 0.4800i 0.6400 - 0.4800i
S1221 =
1.9950 + 0.8400i -1.9100 - 0.6700i 4.0950 + 1.8900i 0.0850 +
0.1700i
S1331 =
2.1000 + 1.0500i -2.0500 - 0.9000i -1.3860 - 0.4520i 0.0500 +
0.1500i
S2332 =
-0.6560 - 0.4320i 0.6640 + 0.4480i -2.5660 - 1.1020i 0.0080 +
0.0160i

Since the algorithm converged in less than 10 iterations, there is no need to repeat with
more iterations.

Example 6.8

Read the example and the longhand solution in the book page 216, then return for the
Matlab solution below (chp6ex8.m):

clear; y12=10-j*20;
y13=10-j*30;
y23=16-j*32;
y33=y13+y23;
V1=1.05+j*0;
format short E
6
iter =0;
S2=-4.0-j*2.5;
P3 = 2;
V2=1+j*0;
Vm3=1.04;
V3=1.04+j*0;
for I=1:10;
iter=iter+1; %***
E2 = V2;
E3=V3;
V2 = (conj(S2)/conj(V2)+y12*V1+y23*V3)/(y12+y23);%***
DV2 = V2-E2;%***
Q3 = -imag(conj(V3)*(y33*V3-y13*V1-y23*V2));%***
S3 = P3 +j*Q3;
Vc3 = (conj(S3)/conj(V3)+y13*V1+y23*V2)/(y13+y23);%***
Vi3 = imag(Vc3);
Vr3= sqrt(Vm3^2 - Vi3^2);
V3 = Vr3 + j*Vi3;%***
DV3=V3-E3;%***
if iter==9
iter
V2, DV2, Q3, Vc3, V3, DV3
elseif iter==10
iter
V2, DV2, Q3, Vc3, V3, DV3
end
end
I12=y12*(V1-V2); I21=-I12;
I13=y13*(V1-V3); I31=-I13;
I23=y23*(V2-V3); I32=-I23;
S12=V1*conj(I12); S21=V2*conj(I21);
S13=V1*conj(I13); S31=V3*conj(I31);
S23=V2*conj(I23); S32=V3*conj(I32);
I1221=[I12,I21]
I1331=[I13,I31]
I2332=[I23,I32]
S1221=[S12, S21 (S12+S13) S12+S21]
S1331=[S13, S31 (S31+S32) S13+S31]
S2332=[S23, S32 (S23+S21) S23+S32]

iter =
9
V2 =
0.9706 - 0.0457i
DV2 =
-8.1775e-007 -6.5578e-006i
Q3 =
1.4616
Vc3 =
1.0400 - 0.0091i
V3 =
1.0400 - 0.0091i
DV3 =
-4.5070e-008 -5.1804e-006i
iter =
10
V2 =
7
0.9706 - 0.0457i
DV2 =
-3.3824e-007 -2.7125e-006i
Q3 =
1.4617
Vc3 =
1.0400 - 0.0091i
V3 =
1.0400 - 0.0091i
DV3 =
-1.8650e-008 -2.1428e-006i
I1221 =
1.7082 - 1.1308i -1.7082 + 1.1308i
I1331 =
0.3720 - 0.2107i -0.3720 + 0.2107i
I2332 =
-2.2828 + 1.6329i 2.2828 - 1.6329i
S1221 =
1.7936 + 1.1874i -1.7096 - 1.0195i 2.1841 + 1.4085i 0.0839 +
0.1679i
S1331 =
0.3906 + 0.2212i -0.3887 - 0.2157i 2.0000 + 1.4618i 0.0018 +
0.0055i
S2332 =
-2.2903 - 1.4805i 2.3888 + 1.6775i -3.9999 - 2.5000i 0.0985 +
0.1969i

As can be seen, after 9 and 10 iterations, the accuracy of the voltages is of the order of
6
10

.

Cover the power flow diagrams for these two examples.

Skip section 6.7 on tap changing transformers.
Chapter 6. Power Flow Analysis (Part IV)


6.8 Power Flow Programs

Read carefully! Note that if tap changing transformers are not used, instead of tap setting
the author uses "1". See for example the last column on page 227. Note that these
"Programs" must be run in a very specific order (see p. 222.)


6.10 Newton-Raphson Power Flow Solution

Due to its quadratic convergence, the N-R method has rapid convergence independent
from system size. Thus, the method usually converges in less than 10 iterations
regardless of system size! However, as the system size (number of equations) increases,
the function evaluations at each iteration increases.

For the typical node shown on the right we can write:
1
n
i ij j
j
I Y V
=
=

. This equation includes bus i under the


summation. Using polar form this equation is written
as:
( )
1
n
i ij i ij j
j
I Y V
=
= +

where
ij ij ij
Y Y = and
j j j
V V = . The conjugate of the complex power at
bus i is given by:
* *
i i i i i
S P jQ V I = = and using the
expression for current we have:
( )
( )
1
n
i i i i ij j ij j
j
P jQ V Y V
=
= +


Solving for real and imaginary parts we have:

( )
1
cos
n
i i j ij ij i j
j
P V V Y
=
= +


and

( )
1
sin
n
i i j ij ij i j
j
Q V V Y
=
= +



If these equations are linearized as before around the point
( ) k
i
and
( ) k
i
V using the
Taylor series expansion around these points we have:
y
i1
y
i2
y
in
y
i0
I
i
V
i
V
1
V
2
V
n
Bus i
2

( ) ( ) ( ) ( )
2 2 2 2
2 2
( )
2
( ) ( ) ( ) ( )
( )
2 2
( ) ( )
( ) ( )
2
2 2
2 2
2
2
( )
( ) ( )
2
|
|
|
|
k k k k
n n
k
k k k k
n n n n
k
n n
n
k k
k k
n
n
k
n
k k
n n
n
P P P P
V V
P
P P P P
V V
P
Q
Q Q
Q Q
V V
Q
Q Q













L L
M O M M O M
M
L L
L
M
L
M O M
L
( )
2
( )
( )
( )
2
( )
( ) ( )
2
|
|
k
k
n
k
k
k
n
k k
n n
n
V
V
Q Q
V V












M
M
M O M
L

Note that
1
P and
1
Q for the slack bus can be computed since for this bus we know
1
V and
1
, hence they are not included above. All quantities are assumed in per unit and the
angles are in radians.

If some bus is a generator bus (P-V bus) then we know P and V for that bus, hence the
equation in Q and V for that bus is not needed. Thus we can eliminate that row and
the corresponding column for the equations above. Thus for each generator bus we have
one less equation (one less unknown) to solve for. If there are n buses, then we have a
maximum of
( )
2 1 n equations. Let us assume we have also 1 m n generator buses,
then the number of equations becomes
( )
2 2 n m equations. Note also that the
Jacobian matrix is evaluated at every iteration step!

The above equation can be put into the more compact form

1 2
3 4
J J P
V J J Q

=




where
1
J ,
( ) ( )
1 1 n n is the derivatives of P with respect to ,
2
J ,
( ) ( )
1 1 n n m is the derivative of P with respect to V ,
3
J ,
( ) ( )
1 1 n m n is
the derivative of Q with respect to , and
4
J ,
( ) ( )
1 1 n m n m is the derivative
of Q with respect to V . All these derivatives are the partial derivatives.

Starting with some initial value for V and and solving these equations for and
V we can proceed to the next iteration thus:

( 1) ( ) ( )
( 1) ( ) ( )
k k k
i i i
k k k
i i i
V V V

+
+
= +
= +

3
We can stop the iteration process when the power residuals are smaller than a pre-
specified value. These are
( ) ( ) k k sch
i i i
P P P = and
( ) ( ) k k sch
i i i
Q Q Q = . These are also
known as the power mismatch.

The general form for the Jacobian elements are as follows. For
1
J the diagonal and off
diagonal elements are:

( )
( )
sin
sin
i
i j ij ij i j
j i
i
i
i j ij ij i j
j
P
V V Y
P
V V Y j i

= +

= +


and the diagonal and off diagonal elements for
2
J are:

( )
( )
2 cos cos
cos
i
i ii ii j ij ij i j
j i
i
i
i ij ij i j
j
P
V Y V Y
V
P
V Y j i
V

= + +

= +


and the diagonal and off diagonal elements for
3
J are:

( )
( )
cos
cos
i
i j ij ij i j
j i
i
i
i j ij ij i j
j
Q
V V Y
Q
V V Y j i

= +

= +


and the diagonal and off diagonal elements for
4
J are:

( )
( )
2 sin sin
sin
i
i ii ii j ij ij i j
j i
i
i
i ij ij i j
j
Q
V Y V Y
V
Q
V Y j i
V

= +

= +


The procedure for the power flow solution using the Newton-Raphson method is as
follows:

1. For load buses where
sch
i
P and
sch
i
Q are specified, the initial voltage magnitudes and
angles are set equal to that of the slack bus (often unity at zero angle, i.e.
(0)
1.0
i
V =
and
(0)
0.0
i
= ). For generator buses (voltage regulated buses) where
i
V and
sch
i
P
are specified, set the voltage angle
i
equal to that of the slack bus (usually zero, i.e.
(0)
0.0
i
= ).
2. For load buses
( ) k
i
P and
( ) k
i
Q are computed using the equations for
i
P and
i
Q derived
earlier. Now compute the power mismatch
( ) ( ) k k sch
i i i
P P P = and
( ) ( ) k k sch
i i i
Q Q Q = .
4
3. For voltage controlled buses (generator buses),
( ) k
i
P is computed directly, hence
( ) ( ) k k sch
i i i
P P P = is computed next.
4. The four elements of the Jacobian matrix are computed.
5. Now the equations involving the Jacobian are solved (in an optimal manner) for the
change in voltage magnitudes and angles.
6. The new voltage magnitudes and angles are found, thus completing one iteration.
7. The process is repeated till the residuals are less than some specified value , i.e.
( ) k
i
P and
( ) k
i
Q for all buses.

The method will converge at a small number of iterations (usually less than 10)
regardless of the size of the system. In large systems the Jacobian matrix is not updated
at each iteration. Instead, it is kept for a few iterations then updated. This saves a lot of
time in the larger power systems.

Go over example 6.10 page 235 in detail.

% Example 6.10 Newton-Raphson method
V = [1.05; 1.0; 1.04];
d = [0; 0; 0];
Ps=[-4; 2.0];
Qs= -2.5;
YB = [ 20-j*50 -10+j*20 -10+j*30
-10+j*20 26-j*52 -16+j*32
-10+j*30 -16+j*32 26-j*62];
Y= abs(YB); t = angle(YB);

iter=0;
pwracur = 0.00025; % Power accuracy
DC = 10; % Set the maximum power residual to a high value
while max(abs(DC)) > pwracur
iter = iter +1

P=[V(2)*V(1)*Y(2,1)*cos(t(2,1)-d(2)+d(1))+V(2)^2*Y(2,2)*cos(t(2,2))+
V(2)*V(3)*Y(2,3)*cos(t(2,3)-d(2)+d(3));
V(3)*V(1)*Y(3,1)*cos(t(3,1)-d(3)+d(1))+V(3)^2*Y(3,3)*cos(t(3,3))+
V(3)*V(2)*Y(3,2)*cos(t(3,2)-d(3)+d(2))];
Q= -V(2)*V(1)*Y(2,1)*sin(t(2,1)-d(2)+d(1))-V(2)^2*Y(2,2)*sin(t(2,2))-
V(2)*V(3)*Y(2,3)*sin(t(2,3)-d(2)+d(3));
J(1,1)=V(2)*V(1)*Y(2,1)*sin(t(2,1)-d(2)+d(1))+
V(2)*V(3)*Y(2,3)*sin(t(2,3)-d(2)+d(3));
J(1,2)=-V(2)*V(3)*Y(2,3)*sin(t(2,3)-d(2)+d(3));
J(1,3)=V(1)*Y(2,1)*cos(t(2,1)-d(2)+d(1))+2*V(2)*Y(2,2)*cos(t(2,2))+
V(3)*Y(2,3)*cos(t(2,3)-d(2)+d(3));
J(2,1)=-V(3)*V(2)*Y(3,2)*sin(t(3,2)-d(3)+d(2));
J(2,2)=V(3)*V(1)*Y(3,1)*sin(t(3,1)-d(3)+d(1))+
V(3)*V(2)*Y(3,2)*sin(t(3,2)-d(3)+d(2));
J(2,3)=V(3)*Y(2,3)*cos(t(3,2)-d(3)+d(2));
J(3,1)=V(2)*V(1)*Y(2,1)*cos(t(2,1)-d(2)+d(1))+
V(2)*V(3)*Y(2,3)*cos(t(2,3)-d(2)+d(3));
J(3,2)=-V(2)*V(3)*Y(2,3)*cos(t(2,3)-d(2)+d(3));
5
J(3,3)=-V(1)*Y(2,1)*sin(t(2,1)-d(2)+d(1))-2*V(2)*Y(2,2)*sin(t(2,2))-
V(3)*Y(2,3)*sin(t(2,3)-d(2)+d(3));
DP = Ps - P;
DQ = Qs - Q;
DC = [DP; DQ]
J
DX = J\DC
d(2) =d(2)+DX(1);
d(3)=d(3) +DX(2);
V(2)= V(2)+DX(3);
V, d, delta =180/pi*d;
end
P1= V(1)^2*Y(1,1)*cos(t(1,1))+V(1)*V(2)*Y(1,2)*cos(t(1,2)-d(1)+d(2))+
V(1)*V(3)*Y(1,3)*cos(t(1,3)-d(1)+d(3))
Q1=-V(1)^2*Y(1,1)*sin(t(1,1))-V(1)*V(2)*Y(1,2)*sin(t(1,2)-d(1)+d(2))-
V(1)*V(3)*Y(1,3)*sin(t(1,3)-d(1)+d(3))
Q3=-V(3)*V(1)*Y(3,1)*sin(t(3,1)-d(3)+d(1))-V(3)*V(2)*Y(3,2)*
sin(t(3,2)-d(3)+d(2))-V(3)^2*Y(3,3)*sin(t(3,3))

iter =
1
DC =
-2.8600
1.4384
-0.2200
J =
54.2800 -33.2800 24.8600
-33.2800 66.0400 -16.6400
-27.1400 16.6400 49.7200
DX =
-0.0453
-0.0077
-0.0265
V =
1.0500
0.9735
1.0400
d =
0
-0.0453
-0.0077
iter =
2
DC =
-0.0992
0.0217
-0.0509
J =
51.7247 -31.7656 21.3026
-32.9816 65.6564 -15.3791
-28.5386 17.4028 48.1036
DX =
-0.0018
-0.0010
-0.0018
V =
1.0500
6
0.9717
1.0400
d =
0
-0.0471
-0.0087
iter =
3
DC =
1.0e-003 *
-0.2166
0.0382
-0.1430
J =
51.5967 -31.6939 21.1474
-32.9339 65.5976 -15.3516
-28.5482 17.3969 47.9549
DX =
1.0e-005 *
-0.3856
-0.2386
-0.4412
V =
1.0500
0.9717
1.0400
d =
0
-0.0471
-0.0087
P1 =
2.1842
Q1 =
1.4085
Q3 =
1.4618

Note that the N-R method took only 3 iterations.

It is a good idea to practice using the "canned" routines to solve this example. See
example 6.11 on page 239 for a sample on using the canned routines.



6.11 Fast Decoupled Power Flow Solution

Much simplification can be made to the Jacobian in order to simplify the power flow
equations. The Jacobian can be simplified to the point where it need not be evaluated
only once instead of on each iteration step. The basic ideas are that the voltages have a
magnitude very near unity, the machine angles are very small and real power is almost a
function of machine angle, while imaginary power is a function of machine voltage.
These statements are approximately correct, but not 100%. Recall the equations
involving the Jacobian are:
7

1 2
3 4
J J P
V J J Q

=




If we assume that real power is only a function of machine angle, then
2
0 J = . If we also
assume that imaginary power is a function of voltage magnitude only, then
3
0 J = . Thus
the equations become:

1
4
0
0
J P
V J Q

=




or

1
4
P
P J
Q
Q J V V
V


= =

= =



Thus the matrix equation is broken into two decoupled equations which are much easier
to solve. Further simplification can be made by making the Jacobian
1
J and
4
J
independent of the iteration process. The diagonal elements of
1
J are described as
follows:

( )
( )
2
1
sin sin
n
i
i j ij ij i j i ii ii
i
i
P
V V Y V Y

= +



It is noted that the first term on the right is
i
Q , thus we have:

( )
2
2
sin
i
i i ii ii
i
i i ii
P
Q V Y
Q V B

=

Where
( )
sin
ii ii ii
B Y = is the imaginary part of the diagonal elements of the bus
admittance matrix.
ii
B is the sum of the susceptance of all the elements connected to bus
i . Typically,
ii i
B Q ? , and
2
i i
V V hence we have:

i
i ii
i
P
V B


The off diagonal elements of
1
J are simplified as follows. Under normal operating
conditions the difference
j i
is so small that we can assume
ij i j ij
+ . Thus
the off diagonal elements of
1
J become

( )
sin
i
i j ij ij
j
i j ij
P
V V Y
V V B

=

Further simplification results from assuming 1
j
V thus

i
i ij
j
P
V B


8
In a similar manner the diagonal elements of
4
J may be written as:

( )
1
sin sin
n
i
i ii ii i j ij ij i j
j
i
Q
V Y V V Y
V

=

= +



where we have used the fact that 1
j
V . The summation term may be replaced by
i
Q
thus we now have
sin
i
i ii ii i
i
Q
V Y Q
V

= +


Similar to what was done earlier, since sin
ii ii ii i
B Y Q = ? , we have:

i
i ii
i
Q
V B
V


And finally, assuming
ij i j ij
+ , the off diagonal terms of
4
J can be written as
follows, :
i
i ij
j
Q
V B
V

= .
Now the equations can be written as

i
i
P
B
V
Q
B V
V

=

Where B and B are the imaginary parts of the bus admittance matrix
bus
Y . Since the
elements of this matrix are constant, they need to be computed only once and need not be
updated at each iteration. These two matrices are triangularized once and used for the
rest of the solution in all the iterations. B is of order 1 n because we remove the first
row and column of the bus admittance matrix (the first bus is the slack bus). Now B is
formed from the bus admittance matrix after deleting the rows and columns that
correspond to generator buses, thus it is of size 1 n m where m is the number of
generator buses. Clearly B is a sub-matrix of B .

Now the equations above can be solved as follows:

[ ]
[ ]
1
1
P
B
V
Q
V B
V

=

This solution is much quicker per iteration than the N-R iteration, but it is not as good in
convergence, hence more iterations are needed. This method is good when a very fast
solution is needed, or as a starting solution for the N-R method.

Go over example 6.12 page 243 in detail, then see the Matlab solution below.

% Example 6.12 Fast decoupled method
V1= 1.05; V2 = 1.0; V3 = 1.04;
9
d1 = 0; d2 = 0; d3=0;
Ps2=-4; Ps3 =2.0;
Qs2= -2.5;
YB = [ 20-j*50 -10+j*20 -10+j*30
-10+j*20 26-j*52 -16+j*32
-10+j*30 -16+j*32 26-j*62];
Y = abs(YB); t=angle(YB);
B =[-52 32; 32 -62]
Binv = inv(B)
iter=0;
pwracur = 0.0003; % Power accuracy
DC = 10; % Set the max of power mismatch to a high value
while max(abs(DC)) > pwracur
iter = iter +1;
P2= V2*V1*Y(2,1)*cos(t(2,1)-d2+d1)+V2^2*Y(2,2)*cos(t(2,2))+
V2*V3*Y(2,3)*cos(t(2,3)-d2+d3);
P3= V3*V1*Y(3,1)*cos(t(3,1)-d3+d1)+V3^2*Y(3,3)*cos(t(3,3))+
V3*V2*Y(3,2)*cos(t(3,2)-d3+d2);
Q2=-V2*V1*Y(2,1)*sin(t(2,1)-d2+d1)-V2^2*Y(2,2)*sin(t(2,2))-
V2*V3*Y(2,3)*sin(t(2,3)-d2+d3);
DP2 = Ps2 - P2; DP2V = DP2/V2;
DP3 = Ps3 - P3; DP3V = DP3/V3;
DQ2 = Qs2 - Q2; DQ2V = DQ2/V2;
DC =[DP2; DP3; DQ2];
Dd = -Binv*[DP2V;DP3V];
DV = -1/B(1,1)*DQ2V;
d2 =d2+Dd(1);
d3 =d3+Dd(2);
V2= V2+DV;
angle2 =180/pi*d2;
angle3 =180/pi*d3;
R = [iter d2 d3 V2 DP2 DP3 DQ2];
disp(R)
end
Q3=-V3*V1*Y(3,1)*sin(t(3,1)-d3+d1)-V3^2*Y(3,3)*sin(t(3,3))-
V3*V2*Y(3,2)*sin(t(3,2)-d3+d2);
P1= V1^2*Y(1,1)*cos(t(1,1))+V1*V2*Y(1,2)*cos(t(1,2)-d1+d2)+
V1*V3*Y(1,3)*cos(t(1,3)-d1+d3);
Q1=-V1^2*Y(1,1)*sin(t(1,1))-V1*V2*Y(1,2)*sin(t(1,2)-d1+d2)-
V1*V3*Y(1,3)*sin(t(1,3)-d1+d3);
S1=P1+j*Q1
Q3

B =
-52 32
32 -62
Binv =
-0.0282 -0.0145
-0.0145 -0.0236

1.0000 -0.0605 -0.0089 0.9958 -2.8600 1.4384 -0.2200
2.0000 -0.0565 -0.0080 0.9653 0.1759 -0.0710 -1.5790
3.0000 -0.0442 -0.0087 0.9657 0.6403 -0.4570 0.0219
4.0000 -0.0448 -0.0090 0.9730 -0.0214 0.0012 0.3652
5.0000 -0.0477 -0.0087 0.9731 -0.1534 0.1129 0.0067
6.0000 -0.0476 -0.0086 0.9714 0.0005 0.0026 -0.0861
7.0000 -0.0469 -0.0087 0.9713 0.0360 -0.0262 -0.0041
10
8.0000 -0.0469 -0.0087 0.9717 0.0009 -0.0014 0.0201
9.0000 -0.0471 -0.0087 0.9718 -0.0084 0.0061 0.0016
10.0000 -0.0471 -0.0087 0.9717 -0.0005 0.0005 -0.0047
11.0000 -0.0471 -0.0087 0.9717 0.0020 -0.0014 -0.0005
12.0000 -0.0471 -0.0087 0.9717 0.0002 -0.0002 0.0011
13.0000 -0.0471 -0.0087 0.9717 -0.0005 0.0003 0.0002
14.0000 -0.0471 -0.0087 0.9717 -0.0001 0.0000 -0.0003
S1 =
2.1843 + 1.4085i
Q3 =
1.4617
Chapter 7
Optimal Dispatch of Generation Part I


7.1 Introduction

When generators are interconnected with various loads, the generation capacity is usually
much larger than the loads, hence the allocation of loads on the generators can be varied.
It is important to reduce the cost of electricity as much as possible, hence the "optimal"
division of load (meaning the least expensive) is desired. The most expensive cost of
generation is the fuel cost. Other expenses include labor, repairs and maintenance.
These are "economical" factors. Other factors (not considered in this presentation)
include such factors as security, stability and aesthetics.


7.2 Nonlinear Function Optimization

Unconstrained Parameter Optimization

The necessary condition for the function
( ) ( )
1 2
, , ,
n
f x f x x x K to have a minimum is
obtained by setting the derivatives 0, 1,2, ,
i
f
i n
x

K . Or using the operator,


0 f where
1 2
, , ,
n
f f f
f
x x x
_



,
K is known as the gradient of f . The term
H associated with second derivative:
2
i j
f
H
x x


results in a symmetric matrix called the
Hessian matrix of the function
( )
f x . Suppose
( )
0 f x where
( )
1 2
, , ,
n
x x x x K is a
local extremum. For this to be a minimum, the Hessian matrix should be positive
definite. This can be checked by finding the eigenvalues of the Hessian matrix, all of
which should be positive at x . In the process of doing this, two functions from Matlab
will be very useful, they are shown below. If x Ab , then the solution is found in
Matlab as \ x A b . Also, the eigenvectors of H are found using the Matlab function
eig(H).

Example 7.1

Find the minimum of
( )
2 2 2
1 2 3 1 2 2 3 1 2 3
2 3 8 16 32 110 f x x x x x x x x x x x + + + + + .
Finding the first derivatives and equate to zero results in three linear algebraic equations
thus:
2

1 2
1
1 2 3
2
2 3
3
2 8 0
4 16 0
6 32 0
f
x x
x
f
x x x
x
f
x x
x

+ +


or:

1
2
3
2 1 0 8
1 4 1 16
0 1 6 32
x
x
x
1 1 1
1 1 1

1 1 1
1 1 1
] ] ]

Using Matlab we have:

A=[2 1 0; 1 4 1; 0 1 6];
b=[8;16;32];
x=A\b
eigenvalues=eig(A)

x =
3.0000
2.0000
5.0000
eigenvalues =
1.5505
4.0000
6.4495

Thus the eigenvalues are all positive, thus x=(3, 2, 5) is a minimum (in this case a global
minimum since there is only one extremum).


7.2.1 Constrained Parameter Optimization:
Equality Constraints

Here we minimize
( )
f x subject to the constraints
( )
0, 1,2, ,
i
g x i k K . Such
problems can be solved using Lagrange multipliers
i
thus we let
1
k
i i
i
f g

L . The
necessary conditions for constrained local minima of L are the following:

1
0
0
k
i
i
i i i i
i
i
f g
x x x
g

L
L

Note that the last equation is the same as the original constraints.

Example 7.2
3

Use the Lagrange multiplier method for solving constrained parameter optimizations to
determine the minimum distance from the origin of the xy-plane to a circle described by
( ) ( )
2 2
8 6 25 x y + . The distance to be minimized is given by
( )
2 2
, f x y x y + .

First we use Matlab to show a graphical display of the two curves from which it is clear
that the answer is at (4, 3). Note there is also a maximum at (12, 9).

wt=0:.01:2*pi;
z =8+j*6+ 5*(cos(wt) + j*sin(wt));
x=0:.01:12; y=6/8*x;
plot(real(z),imag(z), x, y), grid
xlabel('x'), ylabel('y')
axis([0 , 14, 0, 14])

0 2 4 6 8 10 12 14
0
2
4
6
8
10
12
14
x
y


Now we want to minimize
( )
, f x y subject to the constraints described by the circle
equation. We first form the Lagrange function thus:

( ) ( )
2 2
2 2
8 6 25 x y x y
1
+ + +
]
L
The necessary conditions for an extremum are found as
4

( ) ( )
( ) ( )
( ) ( )
2 2
2 2 16 0 or 2 1 16
2 2 12 0 2 1 12
8 6 25 0
x x x
x
y y or y
y
x y

+ +

+ +

L
L
L

Dividing the first two equations yields
3
4
y x . Using this in the third equation yields

2
25
25 75 0
16
x x +
Solving the quadratic gives 4 x and 12 x . Thus the extrema are at the points (4, 3)
with 1 , and (12, 9) with 3 . From the figure it is clear that the first point is the
minimum.

In many situations the equations are not only nonlinear but also cannot be solved in an
analytical manner, thus the solution has to be obtained by iteration. The most common
iteration method is the Newton-Raphson method. This is done below for the above
problem. Solve the first two equations for x and y in terms of :

8
1
6
1
x
y

+

Using these values in the third equation yields:

( ) ( ) ( )
( )
2
2
100 200
, 75 0
1
1
f f x y

+ 1
]
+
+

This equation can be solved by iteration as follows using the Newton-Raphson method.
Using the first order term in a Taylor series expansion we have:

( )
( )
( )
( )
( ) ( ) ( ) 1
and
k
k k k k
k
f
df
d

+
_

,

where
( )
( ) k
f is the residual at
( ) k
given by
( )
( )
( ) ( )
( )
,
k k k
f f x y . If this residual is
zero, the solution is exact, otherwise we would like the residual to satisfy a very small
convergence value. The iteration is started with an estimated value of and continued
till a specified accuracy. Once is known, x and y can be computed from the equations
above. One disadvantage to this method is the need to find the derivative of the function
which in this case is found to be:

( )
( ) ( ) ( )
3 2 3
200 200 200
1 1 1
df
d


+ + +

5
Having a value for
(0)
using the above equation we have
(0)
f

,
. Now we can
compute
(0)
x and
(0)
y hence we have a value for use in:
( ) ( ) ( )
2 2
(0)
( 0) (0)
8 6 25 f x y +
and now we can find
( )
( )
( )
0
(0)
0
f
df
d


_

,
. Thus we have a new value for
(1)
. The
process is repeated till the the error in evaluating f is less than some prescribed value.

The following Matlab program will perform the iteration:

clear;
iter = 0; % Iteration counter
Df = 10; % Error in Df is set to a high value
Lambda = 0.4; % Initial estimated value of Lambda
disp([' Iter Df J DLambda Lambda x
y'])
while abs(Df) >= 0.0001 % Test for convergence
iter = iter + 1; % No. of iterations
x = 8*Lambda/(Lambda + 1);
y = 6*Lambda/(Lambda + 1);
Df = (x- 8)^2 + (y - 6)^2 - 25; % Residual
J = 200*Lambda/(Lambda + 1)^3 - 200/(Lambda + 1)^2; % Derivative
Delambda =-Df/J; % Change in variable
disp([iter, Df, J, Delambda, Lambda, x, y])
Lambda = Lambda + Delambda; % Successive solution
end

Iter Df J DLambda Lambda x y
1.0000 26.0204 -72.8863 0.3570 0.4000 2.2857 1.7143
2.0000 7.3934 -36.8735 0.2005 0.7570 3.4468 2.5851
3.0000 1.0972 -26.6637 0.0411 0.9575 3.9132 2.9349
4.0000 0.0337 -25.0505 0.0013 0.9987 3.9973 2.9980
5.0000 0.0000 -25.0001 0.0000 1.0000 4.0000 3.0000

We repeat the program with an initial value of 2 :

clear;
iter = 0; % Iteration counter
Df = 10; % Error in Df is set to a high value
Lambda = -2; % Initial estimated value of Lambda
disp([' Iter Df J DLambda Lambda x
y'])
while abs(Df) >= 0.0001 % Test for convergence
iter = iter + 1; % No. of iterations
x = 8*Lambda/(Lambda + 1);
y = 6*Lambda/(Lambda + 1);
Df = (x- 8)^2 + (y - 6)^2 - 25; % Residual
J = 200*Lambda/(Lambda + 1)^3 - 200/(Lambda + 1)^2; % Derivative
Delambda =-Df/J; % Change in variable
6
disp([iter, Df, J, Delambda, Lambda, x, y])
Lambda = Lambda + Delambda; % Successive solution
end

Iter Df J DLambda Lambda x y
1.0000 75.0000 200.0000 -0.3750 -2.0000 16.0000 12.0000
2.0000 27.8926 76.9346 -0.3625 -2.3750 13.8182 10.3636
3.0000 8.1227 38.1258 -0.2131 -2.7375 12.6042 9.4531
4.0000 1.2823 26.9480 -0.0476 -2.9506 12.1013 9.0760
5.0000 0.0454 25.0682 -0.0018 -2.9982 12.0036 9.0027
6.0000 0.0001 25.0001 -0.0000 -3.0000 12.0000 9.0000

Now the solution converges to the "maximum" distance at (12, 9).


7.2.2 Constraint Parameter Optimization:
Inequality Constraints

In practice both equality and inequality constraints are needed. The problem is to
minimize the cost function

( )
1 2
, , ,
n
f x x x K
subject to the equality constraints

( )
1 2
, , , 0 1,2, ,
i n
g x x x i k K K
and inequality constraints

( )
1 2
, , , 0 1,2, ,
j n
u x x x j m K K
The Lagrange multipliers are extended to include the inequality constraints using the
multipliers as shown below:

1 1
k m
i i j j
i j
f g u

+ +

L
The resulting necessary conditions for constrained local minima of L are the following:

0 1, ,
0 1, ,
0 1, ,
0 & 0 1, ,
i
i
i
j
j
j j j
i n
x
g i k
u i m
u i m

>
K
K
K
K
L
L
L

Note that the second equation above is the equality constraints and the last two equations
pertain to the inequality constraints. Suppose
( )
1 2
, , ,
n
x x x K is a relative minimum. The
inequality constraint (third equation above) is said to be inactive if strict inequality holds
at
( )
1 2
, , ,
n
x x x K and 0
j
. On the other hand, when strict equality holds, the
7
constraint is active at this point (i.e. if the constraint
( )
1 2
, , , 0
j j n
u x x x K and
0
j
> ). This is known as the Kuhn-Tucker necessary condition.

Example 7.3

Solve example 7.2 with an additional inequality constraint defined below. The problem
is to find the minimum of the function
( )
2 2
, f x y x y + subject to one equality
constraint
( ) ( ) ( )
2 2
, 8 6 25 0 g x y x y + and one inequality constraint
( )
, 2 12 u x y x y + .

The unconstrained cost function is L as shown below:
( ) ( ) ( )
2 2
2 2
8 6 25 2 12 x y x y x y
1
+ + + + +
]
L
The resulting necessary conditions for constrained local minima of L are

( )
( )
( ) ( )
2 2
2 2 8 2 0
2 2 6 0
8 6 25 0
2 12 0
x x
x
y y
y
x y
x y

+ +

+ +

L
L
L
L

Eliminating from the first two equations results in

( )( )
2 4 1 8 0 x y + +
From the fourth condition we have
12 2 y x
Now substituting for y in the previous equation yields

4 4.8
1
x

+

And substituting for x in the fourth condition we get

4 2.4
1
y

+

Substituting for x and y in the third condition (equality constraint) results in an equation
in terms of :

2 2
4 4.8 4 2.4
8 6 25 0
1 1


+ +
_ _
+

+ +
, ,

from which we have the following equation

2
2 0.36 0 + +
Roots of this equation are at 0.2 and 1.8 . Substituting for these values of
in the expressions for x and y, the corresponding extrema are
8

( ) ( )
( ) ( )
, 5,2 for 0.2, 5.6
, 3,6 for 1.8, 12
x y
x y






Using Matlab these are graphically depicted below. Note that the solution is valid above
the line shown in the figure below, thus only two points satisfy the solution for extrema,
one is a minimum, the other a maximum. These are the points where the circle and the
line intersect, which are exactly the two points found above. The graphical
demonstration below further illustrates this example.

%Graphical Demonstration of Example 7.3

wt=0:.01:2*pi;
z =8+j*6+ 5*(cos(wt) + j*sin(wt));
x=0:.01:12; y=6/8*x;
y2=12-2*x;
plot(real(z),imag(z), x, y, x, y2), grid
xlabel('x'), ylabel('y')
axis([0 , 14, 0, 14])


a =[1 2 .36];
Lambda = roots(a)
X=(4*Lambda+4.8)./(1+Lambda)
Y=(4*Lambda+2.4)./(1+Lambda)
F=sqrt(X.^2+Y.^2)
Mindist=min(F)

Lambda =
-1.8000
-0.2000
X =
3.0000
5.0000
Y =
6.0000
2.0000
F =
6.7082
5.3852
Mindist =
5.3852
9
0 2 4 6 8 10 12 14
0
2
4
6
8
10
12
14
x
y


Chapter 7
Optimal Dispatch of Generation Part II


7.3 Operating Cost of a Thermal Plant

The major cost of plant operation is fuel. Other costs may be added to the fuel-cost and
the sum referred to as the fuel cost. Usually the fuel cost is in $/h and is a function of the
power generation in MW. The fuel cost curve is usually assumed to be parabolic of the
form:
2
i i i i i i
C P P = + + . The incremental fuel-cost curve is slope of the fuel-cost
curve defined by 2
i
i i i
i
dC
P
dP
= + . The incremental fuel-cost curve indicates how
expensive it will be to generate the next increment of power. Units:
i
C is in $/h and
i
i
dC
dP

is in $/MWh. Typical curves are shown below.
2

clear
P1=0:1:500;
C1=500+5.3*P1+0.004*P1.^2;
IC1=5.3+2*0.004*P1;
subplot(2,1,1),plot(P1,C1),grid
title('Plot of fuel-cost'),xlabel('Power, MW'),ylabel('$/h')
subplot(2,1,2),plot(P1,IC1),grid
title('Plot of incremental fuel-cost'),
xlabel('Power, MW'),ylabel('$/MWh')
0 100 200 300 400 500
0
1000
2000
3000
4000
5000
Plot of fuel-cost
Power, MW
$/h
0 100 200 300 400 500
5
6
7
8
9
10
Plot of incremental fuel-cost
Power, MW
$ / MWh






3
7.4 Economic Dispatch Neglecting Losses and
Generator Limits

The simplest economic allocation is that of power generators on the same bus (or in
theory using lossless lines). Suppose we have n generators connected to a bus and
supplying a load demand
D
P . We would like to minimize the total cost of generation

( )
1
2
1
ng
t i
i
n
i i i i i
i
C C
P P
=
=
=
= + +


subject to the constraint

1
ng
i D
i
P P
=
=


This situation is shown in the figure to the
right. It is noted that the number of
"dispatchable" generators is ng while the
total number of generators is n. This
means n ng generators are not subject
to economic allocation (they may be on a
flat load, such as the case of nuclear
generators, or they may have reached a
maximum or minimum limit of
generation). As usual we now form the
function L:

1
ng
t D i
i
C P P
=

= +


L
The necessary conditions for a minimum are:

0
0
i
P

L
L

The first condition results in
( ) 0 1 0
t
i
C
P

+ =


Also we know that
1 2 t ng
C C C C = + + + L , therefore

t i
i i
C dC
P dP

= =


Thus the condition for optimal dispatch is
P
1
C
1
P
D
P
ng
C
ng
P
2
C
2
4
1, ,
i
i
dC
i ng
dP
= = K
or
2
i i i
P + =
And the second condition above simply states
1
ng
i D
i
P P
=
=

which is the same as the


constraint stated earlier. In English: Assuming no limits on generation, and no losses,
then the most economical state of operation is when all the plants have the same
incremental cost , and their total generation equals the load demand. Based on this we
now can find for optimal generation. From the equation above we have

2
i
i
i
P

=
The relations given by this equation are called the coordination equations. Using this
equation in the constraint equation we have

1
2
ng
i
D
i
i
P


and solving for we have

1
1
2
1
2
ng
i
D
i
i
ng
i
i
P

=
=
+
=


Knowing the value of we can compute the powers using
2
i
i
i
P

= .
In the case of lossless power systems (i.e. generators on a common bus) it was easy to
analytically solve the optimal dispatch problem. In the general case where losses are
significant and the generators are geographically dispersed, the solution cannot be found
analytically. In such cases an iterative solution is employed. Such a solution is detailed
below for the case of lossless systems (even though this can be performed analytically in
this case, we do it in order to gain experience for the next case when iteration is a must).

We employ the gradient method to solve the optimization problem by iteration. The
constraint equation is written as follows:

( )
1
2
ng
i
D
i
i
f P

= =


Taking the first order terms of the Taylor series expansion about the point
( ) k
gives

( )
( )
( )
( )
( )
k
k k
D
df
f P
d


+ =



or
5

( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
1
ng
k
k D i k
k
D
i
k k k
k
k
i
P P
P f
P
df df df
d d d
P
dP
d


= = =



or

( )
( )
1
2
k
k
i
P


thus the correction is

( ) ( ) ( ) 1 k k k

+
= +
where

( ) ( )
1
ng
k k
D i
i
P P P
=
=


The iteration is continued till
( ) k
P is smaller than a specified accuracy. Of course, as
usual, a starting value
( ) 1
is used to start the iteration. The steps in this process are
outlined below:

Step 1. Assume an initial value for lambda, say
( ) 1
.
Step 2. Using the equation
( )
( ) 1
1
2
i
i
i
P

= compute all the powers for the first iterate.


Step 3. Check if the sum of powers equals the demand, and find the difference thus:
( ) ( ) 1 1
1
ng
D i
i
P P P
=
=

. If this difference is smaller than the specified accuracy, a solution
is reached, if not, continue.
Step 4. Compute the change in lambda:
( )
( ) 1
1
1
2
i
P

and compute the second


estimate for lambda, namely:
( ) ( ) ( ) 2 1 1
= + .

The process is repeated till step 3 indicates convergence




6
Example 7.4

The fuel cost of three thermal plants in $/h are given as:

2
1 1 1
2
2 2 2
2
3 3 3
500 5.3 0.004
400 5.5 0.006
200 5.8 0.009
C P P
C P P
C P P
= + +
= + +
= + +

Where the powers are in MW. The total load
D
P is 800 MW. Neglecting line losses and
generator limits, find the optimal dispatch and the total cost in $/h
(a) by the analytical method
(b) by graphical demonstration
(c) by iteration.

(a) Lambda is computed directly, thus:

1
1
2
1
2
5.3 5.5 5.8
800
0.008 0.012 0.018
8.5 $/MWh
1 1 1
0.008 0.012 0.018
ng
i
D
i
i
ng
i
i
P

=
=
+
=
+ + +
= =
+ +


The powers can be computed directly thus:

( )
( )
( )
1
2
3
8.5 5.3
400
2 0.004
8.5 5.5
250
2 0.006
8.5 5.8
150
2 0.009
P
P
P

= =

= =

= =

It is easy to verify that the incremental costs at the three plants are equal and the sum of
the powers is equal to the power demand.

(b) We can show this graphically using Matlab.

%Graphical Demonstration of Example 7.4
axis([0 450 6.5 10.5]);
lambda=8.5;
P1=250:10:450; P2 = 150:10:350; P3=100:10:250;
IC1= 5.3 + 0.008*P1;
IC2= 5.5 + 0.012*P2;
IC3= 5.8 + 0.018*P3;
Px = 0:100:400;
plot(P1, IC1, P2, IC2, P3, IC3, Px, lambda*ones(1, length(Px)),'-
m'),xlabel('P MW'), ylabel(' $/MWh'), grid

7
0 50 100 150 200 250 300 350 400 450
7
7.5
8
8.5
9
9.5
10
10.5
P MW
$/ MWh


Choosing lambda less than 8.5 would produce less power than the demand, and larger
than 8.5 would produce too much power, hence it is seen that lambda = 8.5 is the solution
and can be found graphically as shown above.

(c) For the iterative solution, we use Matlab thus:

% Iterative solution Using Newton method
alpha =[500; 400; 200];
beta = [5.3; 5.5; 5.8]; gama=[.004; .006; .009];
PD=800;
DelP = 10; % Error in DelP is set to a high value
lambda = 6; % estimated value of Lambda
disp([' Lambda P1 P2 P3 DP grad
Delambda'])
iter = 0; % Iteration counter
while abs(DelP) >= 0.001 % Test for convergence
iter = iter + 1; % No. of iterations
P = (lambda - beta)./(2*gama);
DelP =PD - sum(P); % Residual
J = sum( ones(length(gama), 1)./(2*gama)); % Gradient sum
Delambda = DelP/J; % Change in variable
disp([lambda, P(1), P(2), P(3), DelP, J, Delambda])
8
lambda = lambda + Delambda; % Successive solution
end
totalcost = sum(alpha + beta.*P + gama.*P.^2)

Lambda P1 P2 P3 DP grad Delambda
6.0000 87.5000 41.6667 11.1111 659.7222 263.8889 2.5000
8.5000 400.0000 250.0000 150.0000 0 263.8889 0
totalcost =
6.6825e+003

The Matlab program available to do this is called "dispatch". The program returns the
system lambda and optimal generation dispatch. The program requires as input at least
two specified reserved variables named "cost" and "Pdt" as shown in the example below.
Optionally, the program accepts four other (optional) reserved variables: "mwlimits",
"B", "B0", and "B00". If the optional variables are not specified, the program assumes
they are zero and solves the problem as a lossless problem with no generator limits. The
program called "gencost" will compute the total generation cost. This is shown below:

Example 7.5

Neglecting generator limits and system losses, use "dispatch" to obtain the optimal
dispatch of generation for the thermal plants in Example 7.4.

We use the following Matlab commands:

cost=[500 5.3 0.004
400 5.5 0.006
200 5.8 0.009];
Pdt=800;
dispatch
gencost

Incremental cost of delivered power (system lambda) = 8.500000 $/MWh
Optimal Dispatch of Generation:

400.0000
250.0000
150.0000

Total generation cost = 6682.50 $/h

The use of "mwlimits" and "B B0 B00" will be shown later as we study generator limits
and lossy systems.




Chapter 7
Optimal Dispatch of Generation Part III


7.5 Economic Dispatch Neglecting Losses and
Including Generator Limits

In the real world, generators have limits on their generation, both minimum and
maximum generation. Thus we have the inequality constraints

( ) ( ) min max
1,2, ,
i i i
P P P i ng K
which are in addition to the equality constraint

1
ng
i D
i
P P


Now the Kuhn-Tucker conditions compliment the Lagrangian conditions to include the
inequality constraints as additional terms. The necessary conditions for optimal dispatch
in lossless systems but including generator limits become:

( ) ( )
( )
( )
min max
max
min
for
for
for
i
i i i
i
i
i i
i
i
i
i
i
dC
P P P
dP
dC
P P
dP
dC
P P
dP





The numerical solution is the same as before. Thus for an estimated value of lambda, the
powers are found for each participating generator. If the power is outside the limits of a
generator, that generator is "pegged" at that limit (either Pmin or Pmax), and the
generator is no longer a participating generator in the optimization of dispatched power
(since its power is pegged at either the high or low limit).

Example 7.6

Find the optimal dispatch and the total cost in $/h for the thermal plants of Example 7.4
when the total load is 975 MW with the following generator limits (in MW):

1
2
3
200 450
150 350
100 225
P
P
P




Assume the initial value of
( ) 1
6 . From the coordination equations we compute the
powers as follows:
2

( )
( )
( )
( )
( )
( )
1
1
1
2
1
3
6 5.3
87.5
2 0.004
6 5.5
41.67
2 0.006
6 5.8
11.11
2 0.009
P
P
P




( )
( )
1
975 87.5 41.67 11.11 834.72 P + +
and we compute change in lambda as

( )
( ) ( ) ( )
1
834.72
3.1632
1 1 1
2 0.004 2 0.006 2 0.009

+ +

Therefore the new value of lambda is

( ) 2
6 3.1632 9.1632 +
Continuing for a second iteration we have

( )
( )
( )
( )
( )
( )
2
1
2
2
2
3
9.163 5.3
482.89
2 0.004
9.163 5.5
305.26
2 0.006
9.163 5.8
186.84
2 0.009
P
P
P



and

( )
( )
2
975 482.89 305.26 186.84 0 P + +
Thus the equality constraint is met, but the power of generator No. 1 exceeds its upper
limit. Thus we change it from 482.89 to 450, thus:

( )
( )
2
975 450 305.26 186.84 32.89 P + +
Now we continue, but we have only two generators participating in the optimal dispatch
since the first generator is "pegged" at 450:

( )
( ) ( )
2
32.89
0.2368
1 1
2 0.006 2 0.009

+

and the new value of lambda is

( ) 3
9.1632 0.2368 9.4 +
And the powers are now as follows:
3

( )
( )
( )
( )
( )
3
1
3
2
3
3
450
9.4 5.5
325
2 0.006
9.4 5.8
200
2 0.009
P
P
P



and

( )
( )
3
975 450 325 200 0 P + +
Thus the equality constraint is met and the above is the optimal generation dispatch. The
total generation cost is found using the powers found above thus:

2 2
2
500 5.3 450 0.004 450 400 5.5 325 0.006 325
200 5.8 200 0.009 200 8,236.25 $/h
t
C + + + + +
+ + +


We could solve this using the following Matlab commands:

cost=[500 5.3 0.004
400 5.5 0.006
200 5.8 0.009];
mwlimits=[200 450
150 350
100 225];
Pdt=975;
dispatch
gencost

Incremental cost of delivered power (system lambda) = 9.400000 $/MWh
Optimal Dispatch of Generation:

450.0000
325.0000
200.0000

Total generation cost = 8236.25 $/h


7.6 Economic Dispatch Including Losses

When the loads and generators are in a small geographical region, transmission losses
may be neglected and the optimal dispatch of generation is achieved with all plants
operating at equal incremental productions cost. However, in a large system over long
distances, transmission losses become significant and need to be considered. We shall
use Kron's loss formula which is

0 00
1 1 1
ng ng ng
L i ij j i i
i j i
P PB P B P B

+ +


4
The coefficients
ij
B are called loss coefficients or B-coefficients. It is assumed with
little error that these coefficients are constant (as long as operation is near the value
where these coefficients are computed).

The economic dispatch problem is to minimize the overall generation cost
i
C , which is a
function of the plant outputs

( )
1
2
1
ng
t i
i
n
i i i
i
C C
P P

+ +


Note: If all generators participate in this minimization, then n ng and either equation
above may be used. If only ng n < generators are participating, then the first equation is
used.

This minimization is to be performed subject to the constraint that the total generation
must equal the demand plus the losses thus

1
ng
i D L
i
P P P


satisfying the inequality constraints:

( ) ( ) min max
1, 2, ,
i i i
P P P i ng K
where
( ) min i
P and
( ) max i
P are the minimum and maximum generating limits,
respectively, for plant i.

Using the Lagrange multiplier and adding additional terms to include the inequality
constraints, we have

( ) ( )
( )
( ) ( )
( )
max max min min
1 1 1
ng ng ng
t D L i i i i i i i
i i i
C P P P P P P P

_
+ + + +


,

L
The constraints should be understood to mean that
( ) max
0
i
when
( ) max i i
P P < and
( ) min
0
i
when
( ) min i i
P P > . Thus if the power is not beyond the limits both
( ) min
0
i

and
( ) max
0
i
thus L has only the equality constraint. If a generator's power is beyond
one of the limits, then that limit is the power to which the generator is "pegged", and the
remaining generators participate in the opti mization of the load dispatch. These
statements are clear from the necessary conditions for a minimum which are:
5

( )
( )
( )
( )
max
max
min
min
0
0
0
0
i
i i
i
i i
i
P
P P
P P

L
L
L
L

The last two equations imply that the power should not go beyond its limit, and when it is
within its limits then
( ) ( ) min max
0
i i
and the Kuhn-Tucker function becomes the
same as the Lagrangian one.

The first equation results in
0 1 0
t L
i i
C P
P P

_
+ +


,

Since
1 2 t ng
C C C C + + + L the above equation becomes
1,2, , **
i L
i i
C P
i ng
P P


+

K
The term
L
i
P
P

is known as the incremental transmission loss. The second condition is


given by

1
ng
i D L
i
P P P


which is precisely the equality constraint imposed by the losses and load demand. The
classical form of the previous equation is

1
1,2, ,
1
i
L
i
i
C
i ng
P
P
P

,
K
or
1,2, ,
i
i
i
C
L i ng
P

K
where
i
L is known as the penalty factor of plant i and is given by
1
1
L
i
P
P
_

,
. Thus the
effect of transmission losses is to introduce a penalty factor with a value that depends on
the location of the plant with respect to the loads. In the above equation it is clear that the
6
most economic cost of dispatch is obtained when the incremental cost multiplied by the
corresponding penalty factor are equal for all the participating plants.

The incremental transmission loss can be found from the loss equation and is given by

0
1
2
ng
L
ij j i
i
j
P
B P B
P



and the incremental production cost is given by
2
i
i i i
i
dC
P
dP
+
Using these two equations in the equation marked ** (reproduced below for convenience)
1,2, , **
i L
i i
C P
i ng
P P


+

K
we have

0
1
2 2
ng
i i i ij j i
j
P B P B

+ + +


or (divide by 2 and pull out
ii
from the summation)

0
1,
1
1 *
2
ng
i i
ii i ij j i
j j i
B P B P B



_ _
+ +

, ,


Expanding the above equation for all plants and detailing it into full matrix form we have

1 1
11 12 1 01
1
2 2
2 21 22 2 02
1 2 0
1
1
1
2
1
ng
ng
ng
ng ng
ng ng ngng ng
B B B B
P
P B B B B
P
B B B B






1 1
+
1 1
1
1 1
1
1 1
+
1
1 1

1
1 1
1
1 1
1
]
1 1
+
1 1
] ]
L
L
M
M M O M M
L

or in short matrix form
EP D
To find the optimal dispatch first an estimated guess of , say
( ) 1
is made. The
simultaneous linear equations above are solved using Matlab (P = E\D). Then the
iterative process is continued using the gradient method. Thus, from equation * above we
have
i
P at the kth iteration expressed as

( )
( )
( )
( ) ( )
( )
( )
0
1 2
***
2
k k k
i i ij j
k j i
i
k
i ii
B B P
P
B


and using this equation in the equality constraint gives
7

( )
( )
( )
( ) ( )
( )
( )
( )
0
1 1
1 2
2
k k k
ng ng
i i ij j
k k j i
D i L
k
i i
i ii
B B P
P P P
B



+
+



or

( )
( ) ( ) k k
D L
f P P +
Expanding the left side above in a Taylor series and keeping only first order terms results
in

( )
( )
( )
( ) ( )
k k k
D L
df
f P P
d

_
+ +

,

or

( )
( ) ( )
( )
( )
( )
( )
( )
( )
( )
1
#
ng
k k
D i L
k
i
k
k
k
k
k
i
P P P
df
d
P
df
d
P
dP
d

+

_

,

_

,

_

,


where

( ) ( ) ( )
1
ng
k k k
D i L
i
P P P P

+ + +


and

( )
( )
( )
( )
( )
0
2
1 1
1 2
2
k
k ng ng
i i ii i i ij j
j i
i
k
i i
i ii
B B B P
P
B

_
+ + +

,
+



This last equation is derived by differentiation with respect to of *** above and
summing the result
2
recall that
u v u u v
v v
1 _


1
,
]
.
Now we have

( ) ( ) ( ) 1 k k k

+
+
The process is repeated till
( ) k
P is less than a specified accuracy.

If we simplify the loss equations and use the simpler
8

2
1
ng
L ii i
i
P B P



(where we assume
00
0 and 0
ij
B B ), the solution of the simultaneous equations
simplifies to

( )
( )
( )
( )
##
2
k
k
i
i
k
i ii
P
B

+

which are known as the coordination equations and equation +++ simplifies to

( )
( )
( )
2
1 1
###
2
k
ng ng
i i ii i
k
i i
i ii
P B
B

,
+


The iteration process remains the same using these simpler functions.

Example 7.7

The fuel cost in $/h of three thermal plants of a power system are

2
1 1 1
2
2 2 2
2
3 3 3
200 7.0 0.008
180 6.3 0.009
140 6.8 0.007
C P P
C P P
C P P
+ +
+ +
+ +

where the powers are in MW. The plant outputs are subject to the following limits

1
2
3
10 MW 85 MW
10 MW 80 MW
10 MW 70 MW
P
P
P




For this example, assume the power loss is given by the simplified expression

( ) ( ) ( ) ( )
2 2 2
1 2 3
0.0218 0.0228 0.0179
L pu pu pu pu
P P P P + +
where the loss coefficients are specified per unit on a 100-MVA base. Determine the
optimal dispatch of generation when the total system load is 150 MW.

First we change the power loss equation above to actual units instead of pu:

2 2 2
1 2 3
0.000218 0.000228 0.000179
L
P P P P + +
where the powers are now in MW.

Assume an initial value of
( ) 1
8.0 . From the coordination equations we have:
9

( )
( )
( )
( )
( )
( )
1
1
1
2
1
3
8.0 7.0
51.3136 MW
2 0.008 8.0 0.000218
8.0 6.3
78.5292 MW
2 0.009 8.0 0.000228
8.0 6.8
71.1575 MW
2 0.007 8.0 0.000179
P
P
P


+

The real power loss is

( )
( ) ( ) ( )
2 2 2 1
0.000218 51.3136 0.000228 78.5292 0.000179 71.1575 2.886
L
P + +
Since 150 MW
D
P , the error is (from equation ##)

( )
( )
1
150 2.8864 51.3136 78.5292 71.1517 48.1139 P + + +
and from equation ### :

( )
( ) ( )
( )
1
3
2 2
1
2
0.008 0.000218 7.0 0.009 0.000228 6.3
2 0.008 8.0 0.000218 2 0.009 8.0 0.000228
0.007 0.000179 6.8
152.4924
2 0.007 8.0 0.000179
i
i
P

+ + _
+ +

,
+ +
+


From equation # we have:

( ) 1
48.1139
0.31552
152.4924



Hence the new estimate for is

( ) 2
8.0 0.31552 7.6845
Repeating the above steps it is found at the fourth step that the change in lambda becomes
very small and convergence is declared at

( ) 4
7.6789
And the optimal dispatch for 7.6789 is given by

( )
( )
( )
( )
( )
( )
4
1
4
2
4
3
7.6789 7.0
51.3136 MW
2 0.008 7.6789 0.000218
7.6789 6.3
64.1317 MW
2 0.009 7.6789 0.000228
7.6789 6.8
52.4767 MW
2 0.007 7.6789 0.000179
P
P
P


+

The real power loss is

( )
( ) ( ) ( )
2 2 2 4
0.000218 35.0907 0.000228 64.1317 0.000179 52.4767 1.699
L
P + +
and the total fuel cost is

( ) ( ) ( )
( ) ( ) ( )
2
2 2
200 7.0 35.0907 0.008 35.0907 180 6.3 64.1317
0.009 64.1317 140 6.8 52.4767 0.007 52.4767 1592.65 $/h
t
C + + + + +
+ + +


10
The "dispatch" program can be used to solve example 7.7. The program is designed so
the loss coefficients are specified in pu. The loss coefficients are arranged in a matrix
form with the reserved name "B". The base MVA must be specified in the reserved
variable name "basemva". If the base MVA is not specified, it is set to 100 MVA. The
following Matlab commands are then used to solve example 7.7:

cost=[200 7.0 0.008
180 6.3 0.009
140 6.8 0.007];
mwlimits=[10 85
10 80
10 70];
Pdt=150;
B=[0.0218 0 0
0 0.0228 0
0 0 0.0179];
basemva=100;
dispatch
gencost

Incremental cost of delivered power (system lambda) = 7.678935 $/MWh
Optimal Dispatch of Generation:

35.0907
64.1318
52.4767

Total generation cost = 1592.65 $/h

It is very impressive how much less time it takes to solve example 7.7 using
the Matlab code


Example 7.8

Figure 7.7 page 295 of the text shows the one-line diagram of a 5-bus 3-generator power
system. Given the loss coefficients (pu on a 100 MVA base) are as follows:

[ ]
[ ]
0
00
0.0218 0.0093 0.0028
0.0093 0.0228 0.0017
0.0028 0.0017 0.0179
0.0003 0.0031 0.0015
0.00030523
B
B
B
1
1

1
1
]


Cost functions, generator limits, and total loads are as given in example 7.7. Use
"dispatch" program to find the optimal dispatch of generation.




11
The following Matlab commands are used:

cost=[200 7.0 0.008
180 6.3 0.009
140 6.8 0.007];
mwlimits=[10 85
10 80
10 70];
Pdt=150;
B=[0.0218 0.0093 0.0028
0.0093 0.0228 0.0017
0.0028 0.0017 0.0179];
B0=[0.0003 0.0031 0.0015];
B00=0.00030523;
basemva=100;
dispatch
gencost

Incremental cost of delivered power (system lambda) = 7.767785 $/MWh
Optimal Dispatch of Generation:

33.4701
64.0974
55.1012

Total generation cost = 1599.98 $/h

Chapter 9 Balanced Faults


9.1 Introduction

The most common types of fault are (in order) single-line-to-ground fault, line-to-line
fault, and double-line-to-ground fault. All of these are unbalanced faults. The balanced
(or three phase) fault is the one when all three lines are shorted to ground. It is usually
rare, but can happen. When a fault occurs it is important to isolate it by opening protec-
tive breakers. To properly set the breakers, the magnitude of the fault currents need to be
known. This is the major point of study in this chapter.

The life of a fault can be divided into the following times which occur in sequence:

1. The sub-transient period which lasts for only a few cycles
2. The transient period which lasts for a much longer period (tens of cycles)
3. The steady state period which lasts till a major change in the transmission network
takes place (like a circuit breaker opening or a line failing)

The magnitude of the current in each period needs to be known so as to program breakers
to operate at the right time. To do this in the balanced fault case we need to know how
to construct the bus impedance matrix using the building algorithm which is detailed later
on in the chapter. (It is possible to invert the Ybus matrix, but it is more expedient to
"build" the Zbus directly.

The author presents three programs that are useful in the solution of three phase faults.
These are: Zbus=zbuild(data) and Zbus=zbuildpi(linedata, gendata, yload). The lat-
ter one is compatible with the power-flow input/output files. The third program
symfault(zdata, Zbus,V) is developed for systematic computation of three phase bal-
anced faults of a large interconnected power system.


9.2 Balanced Three-Phase Fault

During a three-phase fault, the impedance of a generator is a time varying quantity. It is
d
X in the sub-transient period (one to four cycles),
d
X in the transient period (about 30
cycles), and the synchronous reactance
d
X after that. Which one to use depends on the
purpose of the study. Keep in mind that the sub-transient currents can be very large due
to the small size of
d
X . Due to symmetry, the three phase currents during a symmetrical
fault can be solved using ordinary circuit theory. In this case we use Thevenin's theorem
to analyze the fault. If the fault has zero impedance to ground, it is called a solid fault or
bolted fault (all three lines shorted to ground with zero impedance).



2

Example 9.1

The one-line diagram of a three bus system is shown below. Each generator is repre-
sented by an emf behind the transient reactance. All impedances are in per unit on a 100
MVA base and for simplicity, resistances are neglected. The following assumptions are
made:

1. Shunt capacitances are neglected and the system is considered on no-load.
2. All generators are running at their
rated voltage and rated frequency with
their emfs in phase.

Determine the fault current, bus voltages,
and the line currents during the fault when
a balanced three-phase fault with imped-
ance 0.16 pu
f
Z = occurs on
(a) Bus 3.
(b) Bus 2.
(c) Bus 1.

Part (a): The fault on bus 3 is simulated
by connecting a fault impedance Z
f
=
j0.16 pu as shown in the figure below.


According to Thevenin's theo-
rem, the change due to the short
circuit is equivalent to that caused
by the added source V
th
= V
3
(0)
with all other sources shorted as
shown on the right, where V
3
(0)
is the pre-fault voltage at bus 3
(where the fault will occur). It is
now easy to solve for the changes
due to the fault. These changes
are then added to the pre-fault
values to get the post-fault values.
The fault current at bus 3 is
( )
( )
3
3
33
0
f
V
I F
Z Z
=
+
where
33
Z is
the Thevenin impedance seen at
the faulted bus. To solve for this,
the delta is first changed to an equivalent Y then the circuit
is simplified as shown next page. Note that all values are pu.
j0.1
j0.1
j0.2
j0.2
j0.4
j0.4
j0.8
1 2
3
j0.2
j0.4
j0.4
j0.8
1 2
3
j0.4
Z
f
= j0.16
j0.2
j0.4
j0.4
j0.8
1
2
3
j0.4
Z
f
= j0.16
V
th
=V
3
(0)
I
3
(F)
3


Note that the delta is not balanced, hence the Y is not either. Simplification above goes
from left to right with the result that Z
33
is equal to j0.34. Now we can compute the fault
current as ( )
( )
3
3
33
0 1
2
0.34 0.16
f
V
I F j
Z Z j j
= = =
+ +
(note: due to the assumptions made
in the statement of the problem,
( ) ( ) ( )
1 2 3
0 0 0 1 pu V V V = = = ). Now we go backward
through the circuit to compute the currents in the original circuit. First the current divides
into two parts through generators 1 and 2 thus: ( )
1 3
0.6
1.2 pu
0.4 0.6
G
j
I I F j
j j
= =
+
and
( )
2 3
0.4
0.8 pu
0.4 0.6
G
j
I I F j
j j
= =
+
. Thus the changes in bus voltages are given by:

( ) ( )
( ) ( )
( )( )
1
2
3
0 0.2 1.2 0.24 pu
0 0.4 0.8 0.32 pu
0.16 2 1.0 0.68 pu
V j j
V j j
V j j
= =
= =
= =

Adding these changes to the pre-fault values we have the post-fault voltages thus:

( ) ( )
( ) ( )
( ) ( )
1 1 1
2 2 2
3 3 3
0 1.0 0.24 0.76 pu
0 1.0 0.32 0.68 pu
0 1.0 0.68 0.32 pu
V F V V
V F V V
V F V V
= + = =
= + = =
= + = =

The short circuit currents in the lines are found as follows:
j0.2
1 2
3
j0.4
Z
f
= j0.16
V
th
I
3
(F)
j0.2
j0.2
j0.1
3
Z
f
= j0.16
V
th
I
3
(F)
j0.1
j0.24
3
Z
f
= j0.16
V
th
I
3
(F)
j0.34
4

( )
( ) ( )
( )
( ) ( )
( )
( ) ( )
1 2
12
12
1 3
13
13
2 3
23
23
0.76 0.68
0.1 pu
0.8
0.76 0.32
1.1pu
0.4
0.68 0.32
0.9 pu
0.4
V F V F
I F j
z j
V F V F
I F j
z j
V F V F
I F j
z j


= = =


= = =


= = =


Part (b). The pre-fault circuit is shown below followed by the Thevenin equivalent then
the reduced circuits. These are shown from left to right:

The Thevenin impedance Z
22
at the fault point is given by j0.24 as shown in the figure on
the right above. The fault current is given by:
( )
( )
2
2
22
0 1.0
2.5 pu
0.24 0.16
f
V
I F j
Z Z j j
= = =
+ +

This current divides between the two generators as follows:

( )
( )
1 2
2 2
0.4
1.0 pu
0.4 0.6
0.6
1.5 pu
0.4 0.6
G
G
j
I I F j
j j
j
I I F j
j j
= =
+
= =
+

Now we find the changes due to the fault as:

( )( )
( ) ( )
( )
1
2
3
0 0.2 1.0 0.2
0 0.4 1.5 0.6
1.0
0.2 0.4 0.4
2
V j j pu
V j j pu
j
V j pu
= =
= =


= =



By superposition we find the post-fault voltages thus:
j0.2
j0.4
j0.4
j0.8
1 2
3
j0.4
Z
f
= j0.16
j0.2
j0.4
j0.4
j0.8
1 2
3
j0.4
Z
f
= j0.16
I
2
(F)
V
th
j0.2
j0.4
1 2
j0.4
Z
f
= j0.16
I
2
(F)
V
th
2
j0.24
Z
f
= j0.16
I
2
(F)
V
th
5

( ) ( )
( ) ( )
( ) ( )
1 1 1
2 2 2
3 3 3
0 1.0 0.2 0.8 pu
0 1.0 0.6 0.4 pu
0 1.0 0.4 0.6 pu
V F V V
V F V V
V F V V
= + = =
= + = =
= + = =

And the short circuit current in the lines are found to be:

( )
( ) ( )
( )
( ) ( )
( )
( ) ( )
1 2
12
12
1 3
13
13
3 2
32
23
0.8 0.4
0.5 pu
0.8
0.8 0.6
0.5 pu
0.4
0.6 0.4
0.5 pu
0.4
V F V F
I F j
z j
V F V F
I F j
z j
V F V F
I F j
z j


= = =


= = =


= = =


Part (c) is similar, and the three currents would be equal to one another but equal to
0.3125 j instead of j0.5. Please go over part (c) carefully

In this example the voltages were assumed equal to 1 pu. For more accurate calculations,
a load flow study is made and the pre-fault voltages are accurately obtained. Also, the
example above did not have any loads at the buses. If there were any loads they would
be represented as constant impedances to ground. This is a good approximation (the
loads change after a fault, but that may be neglected). The general process is summarized
below:

The pre-fault bus voltages are obtained from the results of a load flow analysis.
Loads are converted to constant impedances to ground using the bus voltages.
The faulted network is reduced into a Thevenin equivalent circuit viewed from the
faulted bus. Applying Thevenin's theorem, changes in bus voltages are obtained.
Bus voltages during the fault are obtained by superposition of the prefault bus volt-
ages and the changes in the bus voltages computed in the previous step.
The currents during the fault may now be obtained in all branches of the faulted sys-
tem.












6
9.3 Short-Circuit Capacity (SCC)

The short-circuit capacity of a bus is a measure of the strength of a bus. The SCC is de-
fined as the product of the magnitudes of the rated bus voltage and the fault current. It is
used to determine the size of the bus bar and to size the breaker used at the bus.

From the above definition we have for the short-circuit MVA at bus k:

( )
3
3 10 MVA
Lk k
SCC V I F

=
where the line-to-line voltage is V
Lk
is in kV and I
k
(F) in A. The symmetrical three-
phase current in per unit is given by
( )
( ) 0
k
k
pu
kk
V
I F
X
=
where
( )
0
k
V is the per unit prefault bus voltage and
kk
X is the per unit reactance at the
point of fault. N.B. System resistance is neglected, thus the current computed is on the
pessimistic (larger) side. The base current is given by

3
10
3
B
B
B
S
I
V

=
where
B
S is the base MVA and
B
V is the line-to-line base voltage in kV. Thus the fault
current in ampere is

( ) ( )
( )
3
0
10
3
k k B
pu
k
B
kk B
I F I F I
V
S
X V
=

=

Thus we have for the SCC:

( ) 0
k
B Lk
kk B
V
S V
SCC
X V
=
If we assume the base voltage is equal to the line voltage (i.e.
Lk B
V V = ) then

( ) 0
k B
kk
V S
SCC
X
=
The prefault bus voltage is usually assumed 1.0 pu, thus we have the following approxi-
mate equation for the short-circuit capacity or the short-circuit MVA

B
kk
S
SCC
X




Chapter 9 Balanced Faults, Part II


9.4 Systematic Fault Analysis Using Bus Impedance Matrix

In the previous analysis we employed the Thevenin model and found the Thevenin volt-
age and impedance by means of circuit manipulation. That method works for small
power circuits only: It is impractical for a real large power system. A systematic method
is presented in this section which applies to power systems of any size. The method can
also be programmed on a computer.

Consider a typical bus of an n-bus power system network as shown below. Here we
show buses i and k. The generator shown is a voltage source behind a reactance which
may be , , or
d d d
X X X . Transmission lines are represented by their equivalent model
and all impedances are pu on a common MVA base. A balanced three-phase fault is to
be applied at bus k through an impedance
f
Z . The prefault voltages are obtained from a
power flow solution and represented by the vector
( )
( )
( )
( )
1
0
0
0
0
k
bus
n
V
V
V
V




=




M
M
(1.1)



i
k
S
i
S
k
Z
f
2
Generally the short circuit currents are so large compared to the steady state loads, the
loads may be neglected. However, the loads can be represented as impedances using the
above voltages for example, the load S
i
may be approximated by the impedance

( )
2
*
0
i
i
i
V
Z
S
= (1.2)
The changes in network voltage due to the fault are found from the system with all volt-
ages shorted to ground and the prefault voltage
( )
0
k
V applied as shown in the figure be-
low:
Note the capacitors for the pi equivalent of line ik. Also note the loads are replaced by
the approximation of impedances. The voltage source at i is grounded and its impedance
remains as shown above. The bus voltage changes caused by the fault in this circuit are
represented by the column vector

1
k bus
n
V
V V
V


M
M
(1.3)
From Thevenin's theorem the voltages during the fault are found from

( ) ( )
0
bus bus bus
V F V V = + (1.4)
We also know for the nodal equations that

bus bus bus
I Y V = (1.5)
where
bus
I is the bus current entering the bus and
bus
Y is the bus admittance matrix. The
diagonal elements of each bus is the sum of the admittances connected to that bus, i.e.

0
m
ii ij
j
Y y j i
=
=

(1.6)
i
k
Z
f
I
k
(F)
V
th
=V
k
(0)
Z
k
Z
i
3
The off-diagonal element is equal to the negative of the admittance between the nodes,
i.e.

ij ji ij
Y Y y = = (1.7)
where
ij
y is the actual admittance between nodes i and j. For the Thevenin circuit above,
the nodal equations are

( )
1 11 1 1
1
1
0
0
k n
k k kk kn k
n nk nn n
V Y Y Y
V Y Y Y I F
Y Y Y V




=





L L
M M M M M M M
L L
M M M M M M M
L L
(1.8)
Note that the minus sign is due to the fact the fault current is shown leaving node k. The
above matrix equation can be written as

( )
bus bus bus
I F Y V = (1.9)
which can be solved for the voltage change thus

( )
bus bus bus
V Z I F = (1.10)
where
1
bus bus
Z Y

= is known as the bus impedance matrix (not to be mixed-up with the
impedance matrix in mesh equations of circuit theory!)

Using (1.10) in (1.4) we have:

( ) ( ) ( )
0
bus bus bus bus
V F V Z I F = + (1.11)
Which can be expanded into matrix form:

( )
( )
( )
( )
( )
( )
( )
1 1 11 1 1
1
1
0
0
0
0 0
k n
k k
k kk kn k
n nk nn n n
V F V Z Z Z
V F V
Z Z Z I F
Z Z Z V F V







= +








L L
M M
M M M M M M
L L
M M M M M M M M
L L
(1.12)
Using the kth equation we have:

( ) ( ) ( )
0
k k kk k
V F V Z I F = (1.13)
Also it is clear that

( ) ( )
k f k
V F Z I F = (1.14)
From (1.13) and (1.14) we have:
( )
( ) 0
k
k
kk f
V
I F
Z Z
=
+
(1.15)
Thus we can solve for any bus voltage using (1.12) for row i:

( ) ( ) ( )
0
i i ik k
V F V Z I F = (1.16)
And using (1.15) in (1.16) we have:
( ) ( )
( ) 0
0
k
i i ik
kk f
V
V F V Z
Z Z
=
+
(1.17)
4
Knowing all the bus voltages during a fault we can solve for all the fault currents, in par-
ticular, for line current from bus i to bus j we have:
( )
( ) ( )
i j
ij
ij
V F V F
I F
z

= (1.18)
It is noted that this method works for any bus under fault (i.e. the faulted bus k could be
any one bus of the buses in the system). [Note that
ij
z is the impedance between bus i
and bus j. It is not the i-j-th element of
bus
Z .]

The method suggested above to find the bus impedance matrix by inversion of the bus
admittance matrix is not feasible for very large power systems. An alternate method for
the direct formation (or building) of the matrix
bus
Z will be discussed in the next section.

Example 9.2

A three-phase fault with a fault impedance of
0.16 pu
f
Z j = occurs at bus 3 in the network for Example
9.1. Using the bus impedance matrix method, compute the
fault current, the bus voltages, and the line currents during
the fault.

First the network in Example 9.1 is redrawn to the right us-
ing impedances. By inspection we can find the
bus
Y thus:
8.75 1.25 2.5
1.25 6.25 2.5
2.5 2.5 5
bus
j j j
Y j j j
j j j


=




Note that this is the
bus
Y before the fault occurs. By inver-
sion using Matlab we have:

Ybus = j*[-8.75 1.25 2.5
1.25 -6.25 2.5
2.5 2.50 -5.0];
Zbus = inv(Ybus)

Zbus =
0 + 0.1600i 0 + 0.0800i 0 + 0.1200i
0 + 0.0800i 0 + 0.2400i 0 + 0.1600i
0 + 0.1200i 0 + 0.1600i 0 + 0.3400i

From equation (1.15) we have:

Zf = j*.16;
V0=[1; 1; 1];
I3F = V0(1)/(Zbus(3,3)+Zf)

I3F =
-j5
-j2.5
-j2.5
-j1.25
1 2
3
-j2.5
Z
f
= j0.16
V
th
=V
3
(0)
I
3
(F)
5
0 - 2.0000i

and from (1.16) we have:

VF = V0-I3F*Zbus(:,3)

VF =
0.7600
0.6800
0.3200

Finally from (1.17) we have:

z12 = j*.8; z13 = j*.4; z23 = j*.4;
%
I12 = (VF(1) - VF(2))/z12
I13 = (VF(1) - VF(3))/z13
I23 = (VF(2) - VF(3))/z23

I12 =
0 - 0.1000i
I13 =
0 - 1.1000i
I23 =
0 - 0.9000i

These results are identical to those found earlier in Example 1. Note that the need for the
repeated simplification to find the Thevenin impedance is removed, but a matrix inver-
sion is introduced. This inversion will also be eliminated by the direct building method
of the bus impedance matrix to be discussed in the next section.


Chapter 9 Balanced Faults, Part III


9.5 Algorithm for Formation of the Bus Impedance Matrix

Using this algorithm we shall build the Impedance Matrix
bus
Z starting with a single
element and adding elements one by one till all the elements of the system are accounted
for in the Impedance Matrix. Some definitions from graph theory are introduced below.
The system shown in (a) above has a graph shown in (b). Each bus becomes a node or a
corner in the graph. The datum, or ground bus is the node number 0 in the graph.
Elements in the power system become edges or elements of the graph. For example, the
line from bus 1 to bus 2 whose impedance is j0.8 Ohm becomes the element 4 in the
graph. A tree of a connected graph is a connected subgraph connecting all the nodes
without forming a loop. The elements of a tree are called branches. A graph contains
multiple trees. The number of branches in a tree is always one less than the number of
nodes, i.e.
1 b n = (1.1)
where n is the number of nodes and b the number of branches in the graph. Once a tree
of a graph is determined, the remaining branches are called links. In figure (c) above one
such tree is shown using solid lines and the links are shown with dashed lines. The
collection of links is called a cotree. If e is the total number of elements in a graph, then
the number of links l in a cotree is
1 l e b e n = = + (1.2)
A loop that contains one link is called a basic loop. The number of basic loops is unique:
it is equal to the number of links which is also equal to the number of independent loop
equations the system has. A cut set is a minimal set of elements which when cut will
divide the graph into two connected subgraphs. A fundamental cut set is a cut set which
contains only one branch. The number of fundamental cut sets is unique: it is equal to the
number of branches which is also equal to the number of independent node equations the
system has.

j0.2
j0.4
j0.4
j0.8
1 2
3
j0.4
0
1 2
3
2
1
3
4
5
0
1 2
3
2
1
3
4
5
(a) (b) (c)
2

The bus impedance matrix can be built up starting with a
single element and the process is to add an element at a time
till all the elements of the system are included. Suppose we
have a partial network as shown on the right with a total of m
buses (in addition to the ground or reference bus). The
corresponding equation for this partial network is
m m m
bus bus bus
V Z I = . Here the size of
m
bus
Z is m m . It is
assumed that the network has n buses where n m . Next
we add branches or elements, one by one, till all the
elements of the network are accounted for. The added
element may be a branch or a link described as follows.

Addition of a Branch

A branch can be added in one of two ways: either as a link between two existing nodes,
or as a new branch with a new node. We consider the latter case now (the first case is
considered next). The addition of a new node and branch is shown in the figures below.
First consider (a) above where the new branch is not connected to the reference. The
existing bus impedance matrix is of size m m and after the addition of the branch p q
its size will be
( ) ( )
1 1 m m + + . The impedance of the added branch is
pq
z . The
network equations become:

11 1 12 1 1
1 1
21 22 2 2
2 2
1
2
2
1 2
1 2
|
|
|
|
|
|
|
q
q
pq
p m
p m
p p
p p pp pm
m m
m m m mq p m
q
q q qp q q
m
m q
Z
Z
Z
Z
V
Z Z Z Z Z
Z Z Z Z
V I
Z Z Z Z
V I
V I
Z Z Z Z
V I
Z Z Z Z












=
















L L
L L
M
M M M M M M M
L L
M
M M M M
L
M M
L
M
L L
q
I












(1.3)
Partial
Network
Z
m
bus
1
2
i
m
0
.
.
.
Reference
.
.
.
Partial
Network
Z
m
bus
1
2
p
m
0
.
.
.
Reference
.
.
.
Partial
Network
Z
m
bus
1
2
p
m
0
.
.
.
Reference
.
.
.
(a) (b)
q
q
3
Note: black is the old equation, red is due to the added branch.
The new elements are computed one by one as follows: first compute the elements
qi
Z by
connecting a current source
i
I whose current is 1 (unity) and let all other currents be 0
(zero). Expanding the equations in the matrix above we have:

1 1
2 2
i
i
p pi
m mi
q qi
V Z
V Z
V Z
V Z
V Z
=
=
=
=
=
M
M
(1.4)
From the figure above

q p pq
V V v = (1.5)
where
pq
v is the voltage across the added branch whose impedance is
pq
z , therefore

pq pq pq
v z i = (1.6)
Since there is no current in the added branch, 0
pq
i = , hence 0
pq
v = and (1.5) becomes

q p
V V = (1.7)
And from (1.4) we conclude that

qi pi
Z Z = (1.8)
This means that all the off-diagonal terms in row q are the same as those in row p. Thus
we have found all the red terms except for the term
qq
Z . Now we inject a current at the
qth bus, i.e. 1
q
I = and all other currents set to zero, we get from (1.3)

q qq
V Z = (1.9)
Note that since the current is injected into the node q, we have 1
pq q
i I = = thus from
(1.6) we have

pq pq
v z = (1.10)
And using this in (1.5) we have

q p pq
V V z = + (1.11)
In this case the injected current is at node q, thus setting i q = in (1.4) we have
q qq p pq
V Z and V Z = = hence (1.11) becomes

qq pq pq
Z Z z = + (1.12)

Next we consider the case of figure (b) where the new node and added branch are
connected to the reference node. Clearly this addition is a simple one whose equation is
(by inspection of figure (b) above)

q pq q
V z I = (1.13)


4
Therefore we conclude that all the new elements would be zero except the diagonal
element whose value is
qq pq
Z z = .


Addition of a Link

A link can be added in two ways to two existing nodes as shown below
The impedance is connected either between two existing buses as in (c), or between an
existing bus and ground as in (d). In either case the number of nodes is not changed,
hence the size of the bus impedance matrix must remain the same. First consider (c)
above. Let I
l
be the current in the link pq in the direction shown in the figure. Then

pq p q
z I V V =
l
(1.14)
or
0
q p pq
V V z I + =
l
(1.15)
As seen in the figure, the addition of the link modifies the old currents by changing the
old current
p
I to
( )
p
I I
l
and the old current
( )
to
q q
I I I +
l
. This changes the old
equations as follows

( ) ( )
( ) ( )
( ) ( )
( ) ( )
1 11 1 1 1 1
1 1
1 1
1 1
p p q q m m
p p pp p pq q pm m
q q qp p qq q qm m
m m mp p mq q mm m
V Z I Z I I Z I I Z I
V Z I Z I I Z I I Z I
V Z I Z I I Z I I Z I
V Z I Z I I Z I I Z I
= + + + + + +
= + + + + + +
= + + + + + +
= + + + + + +
l l
l l
l l
l l
L L
M
L L
L L
M
L L
(1.16)
And using the values of and
p q
V V from (1.16) in (1.15) we have one more equation:

( ) ( ) ( )
( ) ( )
1 1
2 0
q p qp pp qq pq
qm pm pq pp qq pq
Z Z I Z Z I Z Z I
Z Z I z Z Z Z I
+ + + + + +
+ + + =
l
L L L
(1.17)
And now combining the above equation into matrix form we have
Partial
Network
Z
m
bus
1
p
m
0
.
.
.
Reference
.
.
.
Partial
Network
Z
m
bus
1
m
0
.
.
.
Reference
.
.
.
(c) (d)
q
p
q
5

11 1 1 1
1 1
1
1
1
1
1
0
p
q
p q m
p pp pq pm
p p
q qp qq qm q q
m
p
m mp m
q
m
m
q m
m m
Z Z Z Z
I V
Z
Z
Z
Z
Z
I Z
Z Z Z
I V
Z Z Z Z I V
Z Z Z Z
Z Z Z Z
I V










=












l
l
l
l
l l l l l ll
L L
M M M M M M M
M M
L L
L L
M M M M M M M
M M
L L
L L
(1.18)
where

i i iq ip
Z Z Z Z = =
l l
(1.19)
and
2
pq pp qq pq
Z z Z Z Z = + +
ll
(1.20)
Now we partition the matrix equation (1.18) above using submatrices:

0
old
bus bus bus
T
Z Z I V
I
Z Z


=



l
ll
(1.21)
Where each part is clear from the colors in (1.18). Expanding the second equation in
(1.21) we have
0
T
bus
Z I Z I = +
ll l
(1.22)
From which we have
0
T
bus
Z I Z I = +
ll l
(1.23)
and using (1.23) in the first equation in (1.21) we have

T
old
bus bus bus
Z Z
V Z I
Z


=


ll
(1.24)
or

new
bus bus bus
V Z I = (1.25)
where

T
new old
bus bus
Z Z
Z Z
Z

=
ll
(1.26)
It is clear that the new bus impedance matrix is of the same size as the old bus impedance
matrix as stated earlier. This is due to the fact that we did not add a node, we added a
link which does not change the number of equations.

In general the bus impedance matrix can be constructed by adding links, nodes and
branches in any order. However, it is best to start by selecting a tree that contains the
elements connected to the reference node. If more than one element connect a node p to
the reference node, only one element can be selected as a branch leaving the other
elements in the cotree. The step by step procedure for building the bus impedance matrix
by adding elements is described below.

6
The case where the added link is between node p and ground will be discussed under rule
3 below.


Rule 1: Addition of a Tree Branch to the Reference

Start with the branches connected to the reference node. Addition of a branch
0 q
z
between node q and the reference node 0 changes the old bus impedance matrix by
adding a row and column as shown below:

0
0
0
old
new bus
bus
q
Z
Z
z

=



(1.27)
Note that the red zeros are vectors of size m (same size as the old bus impedance matrix).


Rule 2: Addition of a Tree Branch from a New Bus to an Old Bus

Addition of a branch
pq
z between a new node q and an existing node p will increase the
size of the old bus impedance matrix by one, from
( )
m m to
( ) ( )
1 1 m m + + . The
change is shown below:

old
bus ip
new
bus
pi pp pq
Z Z
Z
Z Z z

=
+


(1.28)
where
pi
Z is identical to row p of
old
bus
Z and
ip
Z is identical to column p of
old
bus
Z . Note
that by symmetry
T
pi ip
Z Z = . Also,
pp
Z is the diagonal element of the old bus
impedance matrix in the pth position.


Rule 3: Addition of a Cotree Link Between Two Existing Nodes

When a link is added between existing nodes p and q, the size of the old bus impedance
matrix is increased by one as follows

iq ip
old
bus
new
b
q i
u
i p
s
Z Z
Z
Z
Z
Z Z

=

ll
(1.29)
where
T
qi pi
Z Z Z = is the difference between row q and row p of the old bus
impedance matrix and
iq ip
Z Z Z = is the difference between column q and column p
of the old bus impedance matrix. The element Z
ll
is given by
2
pq pp qq pq
Z z Z Z Z = + +
ll
. The new row and column are eliminated using the
equation
7

T
new old
bus bus
Z Z
Z Z
Z

=
ll
. (1.30)
If the added link is between bus p and the reference bus 0, the above results are changed
as follows:

i
old
bus
n
p
pi
ew
bus
Z
Z
Z
Z Z

=

ll
(1.31)
where
T
pi ip
Z Z = is the negative of the pth row (column) of the old bus impedance
matrix and
pq pp
Z z Z = +
ll
.

Example 9.3

Construct the bus impedance matrix for the network of example 9.1 shown again below:
Note the proper tree is chosen in (c) above. We start with the branches connected to the
reference node. First we include branch 1,
10
0.2 z j = between nodes 1 and 0 (ref. node).
According to rule 1 we have
(1)
11 10
0.2
bus
Z Z z j = = = . Next add branch 2,
20
0.4 z j =
between nodes 2 and 0, we now have

11 (2)
22
0 0.2 0
0 0 0.4
bus
Z j
Z
Z j

= =



Note the zeros added and we only have diagonal terms so far. There are no more
elements to ground, so we go to the next element, branch 3. We are adding a new node
(3) and a new branch (3) to an existing node (1). We use rule 2:

11 12 11
(3)
21 22 21
11 12 11 13
0.2 0 0.2
0 0.4 0
0.2 0 0.6
bus
Z Z Z j j
Z Z Z Z j
Z Z Z z l j


= =

+


This completes all the tree branches. Now we go to the cotree and add the links. We add
link (4),
12
0.8 z j = between existing nodes 1 and 2. We have
j0.2
j0.4
j0.4
j0.8
1 2
3
j0.4
0
1 2
3
2
1
3
4
5
0
1 2
3
2
1
3
4
5
(a) (b) (c)
8
11 12 13 12 11
21 22 21 22 21
(4)
11 12 11 13 32 31
44 21 11 22 21 32 31 44
0.2 0 0.2 0.2
0 0.4 0 0.4
0.2 0 0.6 0.2
0.2 0.4 0.2
bus
Z Z Z Z Z j j j
Z Z Z Z Z j j
Z
Z Z Z z Z Z l j j
j j j Z Z Z Z Z Z Z Z


= =
+




where
( )
44 12 11 22 12
2 0.8 0.2 0.4 2 0 1.4 Z z Z Z Z j j j j j = + + = + + = and

[ ]
44
0.2
1
0.4 0.2 0.4 0.2
1.4
0.2
0.0286 0.0571 0.0286
0.0571 0.1143 0.0571
0.0286 0.0571 0.0286
T
j
Z Z
j j j j
Z j
j
j j j
j j j
j j j



=




and from (1.30) we have

(4)
0.2 0 0.2 0.0286 0.0571 0.0286
0 0.4 0 0.0571 0.1143 0.0571
0.2 0 0.6 0.0286 0.0571 0.0286
0.1714 0.0571 0.1714
0.0571 0.2857 0.0571
0.1714 0.0571 0.0571
bus
j j j j j
Z j j j j
l j j j j
j j j
j j j
j j j



=





=




Repeating the same process by adding link 5,
23
0.4 z j = between nodes 2 and 3 we
finally have:

0.16 0.08 0.12
0.08 0.24 0.16
0.12 0.16 0.34
bus
j j j
Z j j j
j j j


=




which is the same as that found by inversion of the
bus
Y matrix obtained earlier in
example 9.2.


Below we repeat the solution using Matlab starting where the links are added:

Z1=[j*0.2 0 j*0.2
0 j*.4 0
j*0.2 0 j*0.6]
DZ1 = [-j*0.2; j*0.4; -j*0.2]
zl1 = j*0.8 + j*0.2 + j*.4
ZX1=1/zl1*DZ1*(-DZ1)'
Z2 = Z1 - ZX1
DZ2 = [Z2(1,3)-Z2(1,2); Z2(2,3)-Z2(2,2); Z2(3,3)-Z2(3,2)]
zl2=j*0.4+Z2(2,2)+Z2(3,3)-2*Z2(2,3)
ZX2=1/zl2*DZ2*(-DZ2)'
Z3=Z2-ZX2

9
Z1 =
0 + 0.2000i 0 0 + 0.2000i
0 0 + 0.4000i 0
0 + 0.2000i 0 0 + 0.6000i
DZ1 =
0 - 0.2000i
0 + 0.4000i
0 - 0.2000i
zl1 =
0 + 1.4000i
ZX1 =
0 + 0.0286i 0 - 0.0571i 0 + 0.0286i
0 - 0.0571i 0 + 0.1143i 0 - 0.0571i
0 + 0.0286i 0 - 0.0571i 0 + 0.0286i
Z2 =
0 + 0.1714i 0 + 0.0571i 0 + 0.1714i
0 + 0.0571i 0 + 0.2857i 0 + 0.0571i
0 + 0.1714i 0 + 0.0571i 0 + 0.5714i
DZ2 =
0 + 0.1143i
0 - 0.2286i
0 + 0.5143i
zl2 =
0 + 1.1429i
ZX2 =
0 + 0.0114i 0 - 0.0229i 0 + 0.0514i
0 - 0.0229i 0 + 0.0457i 0 - 0.1029i
0 + 0.0514i 0 - 0.1029i 0 + 0.2314i
Z3 =
0 + 0.1600i 0 + 0.0800i 0 + 0.1200i
0 + 0.0800i 0 + 0.2400i 0 + 0.1600i
0 + 0.1200i 0 + 0.1600i 0 + 0.3400i


Example 9.4

The bus impedance matrix for the network shown below is found to be
0.183 0.078 0.141
0.078 0.148 0.106
0.141 0.106 0.267
bus
j j j
Z j j j
j j j


=





The line between buses 1 and 3 with impedance
13
0.56 Z j = is removed by the simultaneous
opening of breakers at both ends of the line.
Determine the new bus impedance matrix.

One way of solving this problem is to add a line
between the buses 1 and 3 whose impedance is
the negative of
13
Z to be removed, i.e. we add 0.56 j between nodes 1 and 3. We
proceed using the equation for adding a bus:
j0.35
1
3
2
j0.56 j0.4
j0.5
j0.3
j0.2
10

11 12 13 13 11
21 22 21 23 21
11 12 11 13 33 31
31 11 32 21 33 31 44
bus
Z Z Z Z Z
Z Z Z Z Z
Z
Z Z Z z Z Z
Z Z Z Z Z Z Z


=
+




where
44 13 11 33 13
2 Z z Z Z Z = + + . Also note that Z is the last column of the matrix
less the element
44
Z . Now it is easy to use Matlab to solve the problem:

Z = [ j*0.183 j*0.078 j*0.141
j*0.078 j*0.148 j*0.106
j*0.141 j*0.106 j*0.267]
DZ = Z(:, 3)-Z(:,1)
Z44 = -j*0.56+Z(1,1)+Z(3,3)-2*Z(1,3)
DZZ =DZ*DZ.'/Z44
Zbus = Z - DZZ

Z =
0 + 0.1830i 0 + 0.0780i 0 + 0.1410i
0 + 0.0780i 0 + 0.1480i 0 + 0.1060i
0 + 0.1410i 0 + 0.1060i 0 + 0.2670i
DZ =
0 - 0.0420i
0 + 0.0280i
0 + 0.1260i
Z44 =
0 - 0.3920i
DZZ =
0 - 0.0045i 0 + 0.0030i 0 + 0.0135i
0 + 0.0030i 0 - 0.0020i 0 - 0.0090i
0 + 0.0135i 0 - 0.0090i 0 - 0.0405i
Zbus =
0 + 0.1875i 0 + 0.0750i 0 + 0.1275i
0 + 0.0750i 0 + 0.1500i 0 + 0.1150i
0 + 0.1275i 0 + 0.1150i 0 + 0.3075i



9.6 Zbuild and Symfault Programs

These two functions are developed for the formation of the bus impedance matrix. The
first zbuild is
Zbus = zbuild(zdata)
zdata is an 4 e matrix containing the impedance data of an e-element network.
Columns 1 and 2 contain the element bus numbers and columns 3 and 4 contain the
element resistance and reactance respectively, in pu. Bus number 0 to generator buses
contain generator impedances. These may be the subtransient, transient, or synchronous
reactances. Also any other loads and capacitors to ground may be included in this matrix.

The second sbuild function for the formation of the bus impedance matrix is
zbus = zbuildpi(linedata, gendata, yload)
11
which is compatible with the power flow programs. The arguments are
linedata this is the data required for the power flow solution. Columns 1 and 2
contain the bus numbers, columns 3 through five contain the resistance, reactance and
half the susceptance for the line all in pu on the same MVA base. The last column
number 6 is for tap settings of transformers, if a line, the enter 1 (one).
gendata is where the generator data is entered. This is an 4
g
n matrix, where each
row contains bus 0, generator bus number, resistance and reactance for each generator.
yload is optional. It contains the results of a load flow solution (bus number and load
impedance) as generated by one of the programs lfgauss, lfnewton, or
decouple. yload is automatically generated following application of one of these
programs.

These two programs (zbuild and zbuildpi) obtain the bus impedance matrix using
the algorithms developed in the chapter. First a tree is selected which contains elements
to the reference node. Then the remaining branches of the tree are connected, and finally
the cotree links are added.

symfault(zdata, Zbus, V) was developed for three phase balanced fault studies.
The arguments zdata and Zbus are required while V is optional. If V is not included, it
is assumed that all pre-fault voltages are 1.0 per unit. If V is included, then it is an array
describing the bus numbers and the complex bus voltage. Again, this data is
automatically generated following a load flow study.
symfault will prompt the user to supply the faulted bus number and the fault
impedance. The program computes the fault current and the bus voltages and line
currents during the fault.

Go through examples 9.5 though 9.9 using the Matlab window.





Chapter 10: Symmetrical Components and
Unbalanced Faults


10.1 Introduction

When an unbalanced three-phase fault occurs, we can solve the three-phase circuit using
ordinary circuit theory. This is much more numerically complicated than the single-
phase circuit normally used in balanced three phase circuits. The degree of difficulty
increases with the third power of the system size. For this reason, it is apparent that if we
were to solve three different single-phase circuits, it would be numerically simpler than
solving the one three-phase circuit in one set of equations.

The purpose of this chapter is to break up the large three-phase circuit into three circuits,
each one third the size of the whole system. Next, we solve the three components
individually, and then combine the results to obtain the total system response. The two
programs abc2sc and sc2abc will transform the system representation from the
normal "abc" form to the "sc" or symmetrical components form (and the other way
around.) Three other programs are developed, these are:

Line to ground fault studies: lgfault(zdata0, zbus0, zdata1, zbus1,
zdata2, zbus2, V)

Line to line fault studies: llfault(zdata1, zbus1, zdata2, zbus2, V)

Double line to ground fault studies: dlgfault(zdata0, zbus0, zdata1,
zbus1, zdata2, zbus2, V)

As can be seen above, these programs use the three components which are developed
first, namely the "0", "1" and "2" components. These are often referred to as the zero-
sequence, positive-sequence and negative-sequence components, respectively. Some
other books use "0", " + " and " " for these same components.


10.2 Fundamentals of Symmetrical Components

It was Fortescue in 1918 who developed the idea of breaking up
asymmetrical three-phase voltages and currents into three sets of
symmetrical components. These three basic components are:
(1) The positive sequence currents and voltages (known also as
the "abc" and often denoted by the superscript "1" or " + ",)
shown on the right. Note this is the regular three-phase
balanced phasors we have been dealing with from the start
of this course.

a
1

b
1

c
1
2
(2) The negative sequence currents and voltages (known also
as the "acb" and often denoted by the superscript "2" or
" ".) These are the regular balanced negative sequence
components we have seen previously. Note the sequence
of the phasors is the opposite direction from the positive
sequence (acb instead of abc.)
(3) The zero sequence components of currents and voltages
(often denoted by the superscript "0".) These components are
shown on the right. Note that these zero sequence phasors are all
in-phase and equal in magnitude.

We will need the vector 1 120 a
o
, which is a unit vector at an angle of
120 degrees. It is easy to see that
2 *
1 240 1 120 a a
o o
and
3
1 a . It is also
clear that
2
1 0 a a + + .

Positive sequence:
1 1 1
1 1 2 1
1 1 1
0
240
120
a a a
b a a
c a a
I I I
I I a I
I I aI



o
o
o


Negative sequence:
2 2 2
2 2 2
2 2 2 2
0
120
240
a a a
b a a
c a a
I I I
I I aI
I I a I



o
o
o


Zero sequence:
0 0 0
a b c
I I I

Now consider a three-phase unbalanced set of currents
a
I ,
b
I , and
c
I as shown in the figure to the right. Note that
the vectors are arbitrary in length and direction. The same
argument would hold if these were voltages instead of
currents. These "abc" currents can be expressed in terms
of the "sc" currents as follows:

0 1 2
0 1 2
0 1 2
a a a a
b b b b
c c c c
I I I I
I I I I
I I I I
+ +
+ +
+ +


And we know that all the currents can be expressed in terms of the "a" components, thus:

0 1 2
0 2 1 2
0 1 2 2
a a a a
b a a a
c a a a
I I I I
I I a I aI
I I aI a I
+ +
+ +
+ +

a
2

c
2

b
2
I
a
0
I
b
0
I
c
0
I
a
I
b
I
c
3
Or in matrix form:

0
2 1
2 2
1 1 1
1
1
a a
b a
c a
I I
I a a I
I a a I
_
_ _





, ,
,

Which can be written as a single matrix equation as:
012 abc
a
I AI , where A is known as
the symmetrical components transformation matrix, which transforms the currents
abc
I
into component currents
012
a
I and is given by:
2
2
1 1 1
1
1
A a a
a a
_



,
. Note that
T
A A , thus
the A-matrix is symmetric.
Using matrix inversion we have:
012 1 abc
a
I A I

and the inverse matrix is given by:



1 2
2
1 1 1
1
1
3
1
A a a
a a



,

And since
2 *
a a , we see that
1 *
1
3
A A

. Using the above equation it is seen that:



0
1 2
2 2
1 1 1
1
1
3
1
a a
a b
a c
I I
I a a I
I a a I
_
_ _





, ,
,

Or in detail:

( )
( )
( )
0
1 2
2 2
1
3
1
3
1
3
a a b c
a a b c
a a b c
I I I I
I I aI a I
I I a I aI
+ +
+ +
+ +

It is noted that
0
a
I is the sum of the three "abc" components, thus if the abc components
are balanced, their sum is zero, hence the zero-sequence component is also zero.

The exact similar expressions are valid for voltages
abc
V and
012
a
V . These equations need
not be repeated here.

Often it is understood that the quantities with superscript "0", "1", or "2" will have
subscript "a" and the "a" is omitted. Thus instead of
0
a
I we use
0
I and instead of
1
a
I we
use
1
I and so on

Next we investigate the three phase complex power. We know that

( )
*
3
T
abc abc
S V I


4
The equation for complex power may be expressed in terms of symmetrical components
thus:

( )
( ) ( )
*
012 012
3
T
a a
S AV AI


And dropping the subscript "a" these may be written as:

( )
( ) ( )
*
012 012
3
*
012 * 012
T
T
T
S AV AI
V A A I


Keeping in mind that "A" is symmetric and
*
3
T
A A , we have:

( ) ( )
*
012 012
3
* * *
0 0 1 1 2 2
3
3 3 3
T
S V I
V I V I V I


+ +

The last equation shows that the total unbalanced three-phase power is equal to the sum
of the three symmetrical component powers. Note that in the above equations the
subscript "a" was dropped from all those terms having superscript "012," "0," "1," or "2."

Please read about the functions defined on page 404: sctm, phasor, abc2sc,
sc2abc, zabc2sc, rec2pol, and pol2rec.

Example 10.1
Obtain the symmetrical components of a set of unbalanced currents 1.6 25
a
I
o
,
1.0 180
b
I
o
, and 0.9 132
c
I
o
.

The Matlab program follows:

Iabc = [1.6 25
1.0 180
0.9 132];
I012 = abc2sc(Iabc); % Symmetrical components of phase a
I012p= rec2pol(I012) % Converts rectangular phasors into polar form

I012p =
0.4512 96.4529
0.9435 -0.0550
0.6024 22.3157
5
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
a-b-c set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Zero-sequence set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Positive-sequence set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Negative-sequence set




Example 10.2

The symmetrical components of a set of unbalanced three-phase voltages are
0
0.6 90 V
o
,
1
1.0 30 V
o
, and
2
0.8 30 V
o
. Obtain the original unbalanced
phasors.

The Matlab program follows:

V012 = [.6 90
1 30
0.8 -30];
Vabc = sc2abc(V012); % Unbalanced phasors from symmetrical components
Vabcp= rec2pol(Vabc) % Converts rectangular phasors into polar form

Vabcp =
1.7088 24.1825
0.4000 90.0000
1.7088 155.8175
6
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
a-b-c set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Zero-sequence set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Positive-sequence set
-1 0 1
-1.5
-1
-0.5
0
0.5
1
1.5
Negative-sequence set



10.3 Sequence Impedances

Each of the sequence currents (or voltages) will encounter a different impedance in the
network. In other words, each network component will present a different impedance as
seen be each of the sequences. The network components are: loads, transmission lines,
generators and transformers. We shall consider the impedances each of these
components presents.
1
Z is the usual impedance for positive sequence currents,
2
Z is
impedance for negative sequence currents, and
0
Z is the impedance for zero sequence
currents.

10.3.1 Sequence Impedances of
Y-Connected Loads

The balanced three-phase Y-load is shown in
the figure to the right. There is a mutual
coupling between the three loads making up
the Y-load. The return current through the
neutral goes through an impedance
n
Z . The
"abc" voltages to ground are:

a s a m b m c n n
b m a s b m c n n
c m a m b s c n n
V Z I Z I Z I Z I
V Z I Z I Z I Z I
V Z I Z I Z I Z I
+ + +
+ + +
+ + +


Z
m
Z
m
Z
m
I
n
Z
n
V
a
V
b
V
c
- - -
7
From Kirchhoff's current law, we have
n a b c
I I I I + + . Using this equation above, we
have:

a s n m n m n a
b m n s n m n b
c m n m n s n c
V Z Z Z Z Z Z I
V Z Z Z Z Z Z I
V Z Z Z Z Z Z I
+ + +
_ _ _

+ + +


+ + +
, , ,


Or in compact form:
abc abc abc
V Z I . Using symmetrical components (recall that
012
a
I is
denoted by
012
I for simplicity,) this equation is written as:
012 012 abc
AV Z AI and solving
for the voltage we have:

012 1 012
012 012
abc
V A Z AI
Z I


Where:

012 1 abc
Z A Z A



Using previously derived matrices we have:

012 2 2
2 2
1 1 1 1 1 1
1
1 1
3
1 1
s n m n m n
m n s n m n
m n m n s n
Z Z Z Z Z Z
Z a a Z Z Z Z Z Z a a
a a Z Z Z Z Z Z a a
+ + +
_ _ _

+ + +


+ + +
, , ,

Performing the above matrix multiplications, we have:

012
3 2 0 0
0 0
0 0
s n m
s m
s m
Z Z Z
Z Z Z
Z Z
+ +
_



,

And if there is no mutual coupling between the legs of the Y-load, we have:

012
3 0 0
0 0
0 0
s n
s
s
Z Z
Z Z
Z
+
_



,

N.B. The impedance matrix has non-zero elements only on the diagonal, therefore for
balanced load the three sequences are independent. This means currents in one sequence
will produce voltages in that sequence only! This is a very important property since it
allows us to analyze each sequence network on a per-phase basis.


10.3.2 Sequence Impedances of Transmission Lines

The transmission lines present the same geometry to the positive as well as the negative
sequence currents, thus for lines we have
1 2
Z Z . However, the zero-sequence
impedance is much larger (in general) and is given by:
0 1
3
n
X X X + where
2 0.2ln m /km
n n
D
X f
D

_


,
. Here
n
D is the distance between the line and the
8
neutral (usually earth or ground,) and D is the distance between the three lines (assuming
they are equidistant.)


10.3.3 Sequence Impedances of Synchronous Machine

Since the machine has rotation, the inductances seen will differ depending on the
direction of rotation of the armature currents. Thus the negative sequence currents will
appear to have twice line frequency. However, zero sequence currents will rotate with
the windings, thus here we would see only the leakages. As an approximation, the
sequence impedances of synchronous machines are approximated as follows:

Positive Sequence:
1
or or
d d d
X X X X , depending on analysis (sub-transient,
transient or steady state.)

Negative Sequence:
2
d
X X ; , this is approximately correct.

Zero Sequence:
0
X X
l
; , where X
l
is the leakage reactance.


10.3.4 Sequence Impedances of Transformers

We will study the per-phase parameters of the equivalent circuit of the transformer. The
shunt (magnetization and core losses representation,) are usually neglected since they
have a high impedance and they represent less than 1% of the rated transformer power.
Also, since the transformer has no rotating fields or components, all the sequence
impedances are the same (though sometimes the zero sequence current does not have a
path to flow.) Hence we have:
0 1 2
Z Z Z Z
l
.

In a balanced delta connection and also in a 3-wire Y-connection, there will be no zero
sequence currents outside the transformer; hence, these impedances will be represented
with an open circuit (infinite zero-sequence impedance.) Recall that it is the convention
to connect Y and Y transformers so that the high voltage side will lead the low
voltage side by 30
o
(this is for the normal, or positive-sequence voltages.) For the
negative-sequence voltages, this phase shift would be 30
o
. The zero-sequence
impedance will depend on the connection of the two sides of the transformer. These are
shown schematically below, for detailed diagrams, see page 413 of the book.


Four-wire Ys:




g
a a'
9


Four-wire Y-Y:





Four-wire Y-Delta:





Y-Delta:





Delta-Delta:



N.B. The neutral (fourth wire) impedance to ground is very important since it appears
with a factor of "3" times
n
Z . (This is due to the fact that if
0
I flows, then the neutral
current will be
0
3
n
I I .)


Example 10.3

A balanced three phase voltage of
100-V line to neutral is applied to a
balanced Y-connected load with
ungrounded neutral. The three-
phase load consists of three
mutually coupled reactances. Each
phase has a series reactance of
12
s
Z j , and the mutual
coupling between phases is
4
m
Z j .
(a) Determine the line currents
by mesh analysis without using symmetrical components.
(b) Determine the line currents using symmetrical components.

g
a a'
g
a a'
g
a a'
g
a a'
Z
m
Z
m
Z
m
V
a
V
b
V
c
- - -
10
(a) Using KVL to the two meshes we have:

/ 6
/ 2
s a m b s b m a a b L
s a m c s c m b b c L
Z I Z I Z I Z I V V V
Z I Z I Z I Z I V V V

+
+

And from KCL we have:
0
a b c
I I I + +
Writing these three equations in matrix form:

( ) ( )
( ) ( )
0 / 6
0 / 2
1 1 1 0
s m s m a L
s m s m b L
c
Z Z Z Z I V
Z Z Z Z I V
I

_ _ _







,
, ,

Or in compact form:
abc
mesh mesh
Z I V . Solving this equation using matrix inversion:

1 abc
mesh mesh
I Z V


The following Matlab program performs these operations:

disp('(a) Solution by mesh analysis:');
Zs=j*12; Zm=j*4; Va=100; VL=sqrt(3)*Va;
Z=[(Zs-Zm) -(Zs-Zm) 0
0 (Zs-Zm) -(Zs-Zm)
1 1 1];
V=[VL*cos(pi/6)+j*VL*sin(pi/6)
VL*cos(-pi/2)+j*VL*sin(-pi/2)
0] ;
Y=inv(Z);
Iabc=Y*V;
Iabcp=[abs(Iabc), angle(Iabc)*180/pi]

12.5000 (a) Solution by mesh analysis:
12.5001 Iabcp =
12.5002 12.5000 -90.0000
12.5003 12.5000 150.0000
12.5004 12.5000 30.0000

(b) Using symmetrical components we have:

012 012 012
V Z I
where:

012
0
0
a
V V
_



,

and from past equations (page 7 with no
n
Z ,) we have:

012
2 0 0
0 0
0 0
s m
s m
s m
Z Z
Z Z Z
Z Z
+
_



,

And now we have:
1
012 012 012
I Z V

1
]
. This is done using the Matlab program below:

11
disp('(b) Solution by symmetrical components method:');
Z012=[Zs+2*Zm 0 0
0 Zs-Zm 0
0 0 Zs-Zm];
V012=[0; Va; 0];
I012=inv(Z012)*V012;
a=cos(2*pi/3)+ j*sin(2*pi/3);
A=[1 1 1; 1 a^2 a; 1 a a^2];
Iabc=A*I012;
Iabcp=[abs(Iabc), angle(Iabc)*180/pi]

(b) Solution by symmetrical components method:
Iabcp =
12.5000 -90.0000
12.5000 150.0000
12.5000 30.0000

Which is identical with the answer obtained in part (a).

Example 10.4:

A three-phase unbalanced source with the following phase to neutral voltages

200 25
100 155
80 100
abc
V
_



,
o
o
o

is applied to the circuit shown. The load series impedance per phase is 8 24
s
Z j + and
the mutual impedance between phases is 4
m
Z j . The load and source neutrals are
solidly grounded. Determine:
(a) The load sequence impedance matrix
012 1 abc
Z A Z A

.
(b) The symmetrical components of voltage.
(c) The symmetrical components of current.
(d) The load phase currents.
(e) The complex power delivered to the load in terms of symmetrical components.
(f) The complex power delivered to the load in terms of the power per phase.

The following Matlab program is the solution:

clear
Vabc= [200 25
100 -155
80 100];
Zabc=[8+j*24 j*4 j*4
j*4 8+j*24 j*4
j*4 j*4 8+j*24];
Z012=zabc2sc(Zabc)
V012=abc2sc(Vabc);
V012p=rec2pol(V012)
I012=inv(Z012)*V012;
I012p=rec2pol(I012)
Iabc=sc2abc(I012);
12
Iabcp=rec2pol(Iabc)
S3ph_sc=3*(V012.')*conj(I012)
Vabcr=Vabc(:, 1).*(cos(pi/180*Vabc(:, 2))+j*sin(pi/180*Vabc(:,2)));
S3ph=(Vabcr.')*conj(Iabc)

Z012 =
8.0000 +32.0000i 0 + 0.0000i 0 + 0.0000i
0.0000 + 0.0000i 8.0000 +20.0000i -0.0000 + 0.0000i
0 + 0.0000i 0.0000 - 0.0000i 8.0000 +20.0000i
V012p =
47.7739 57.6268
112.7841 -0.0331
61.6231 45.8825
I012p =
1.4484 -18.3369
5.2359 -68.2317
2.8608 -22.3161
Iabcp =
8.7507 -47.0439
5.2292 143.2451
3.0280 39.0675
S3ph_sc =
9.0471e+002 +2.3373e+003i
S3ph =
9.0471e+002 +2.3373e+003i
-5 0 5
-6
-4
-2
0
2
4
6
a-b-c set
-5 0 5
-6
-4
-2
0
2
4
6
Zero-sequence set
-5 0 5
-6
-4
-2
0
2
4
6
Positive-sequence set
-5 0 5
-6
-4
-2
0
2
4
6
Negative-sequence set



Chapter 10: Symmetrical Components and
Unbalanced Faults, Part II


10.4 Sequence Networks of a Loaded Generator

In the figure to the right is a generator supplying a
three-phase load with neutral connected through
impedance
n
Z to ground. The generator generates
three-phase balanced voltages defined as:

2
1
abc
a
E a E
a
_



,

The terminal voltage of the generator is found
using Kirchoff's voltage law:

a a s a n n
b b s b n n
c c s c n n
V E Z I Z I
V E Z I Z I
V E Z I Z I




And since
n a b c
I I I I + + , we have:

a a s n n n a
b b n s n n b
c c n n s n c
V E Z Z Z Z I
V E Z Z Z Z I
V E Z Z Z Z I
+
_ _ _ _

+


+
, , , ,

Or in compact form:
abc abc abc abc
V E Z I . This equation can be transformed to the
"012" or symmetrical component form thus:

012 012 012 abc
a a a
AV AE Z AI
Multiplying by
1
A

we have:

012 012 1 012
012 012 012
abc
a a a
a a
V E A Z AI
E Z I




Where

012 1
2 2
2 2
1 1 1 1 1 1
1
1 1
3
1 1
abc
s n n n
n s n n
n n s n
Z A Z A
Z Z Z Z
a a Z Z Z Z a a
a a Z Z Z Z a a

+
_ _ _

+


+
, , ,

Performing the above matrix multiplication we have:

0
012 1
2
3 0 0 0 0
0 0 0 0
0 0 0 0
s n
s
s
Z Z Z
Z Z Z
Z Z
_ +
_







,
,

E
a
E
b
E
c
Z
s
Z
s
Z
s
Z
n
I
a
I
b
I
c
I
n
+
+
+
-
- -
V
a
V
c
V
b
2
It is important to note that the matrices above are diagonal, and the equations are
completely decoupled. Expanding into three voltage equations we have:

0 0 0
1 1 1
2 2 2
0
( )
0
a a
a a a
a a
V Z I
V E Z I
V Z I

++


These equations
represent three separate
circuits as shown to the
right, one for the
positive-sequence, one
for the negative-
sequence, and one for
the zero-sequence
networks.

The following important observations are made:

The three sequences are independent.
The positive-sequence network is the same one used in the one-line diagram for
studying balanced three-phase currents and voltages.
Only the positive sequence-network has a voltage source. Thus the positive-
sequence voltage will generate only positive-sequence currents.
There is no voltage source in the negative- and zero-sequence networks.
Negative- and zero-sequence currents cause only negative- and zero-sequence
voltages.
The neutral of the system is the reference point for the positive- and negative-
sequence networks. The ground is the reference point for the zero-sequence;
hence the zero-sequence current cannot flow unless there is a connection to
ground.
The grounding impedance is reflected in the zero sequence network as 3
n
Z .
Each of the three sequence networks can be solved separately on a per phase
basis. The phase currents and voltages can then be found by superposition, by
adding their symmetrical components of current and voltage respectively.

10.5 Single Line-To-Ground Fault

This is the most common fault on a three-phase system. It is illustrated using the
following simple network:
E
a
+
-
V
1
a
+
-
V
2
a
+
-
V
0
a
+
-
I
1
a
Z
1
I
2
a
Z
2
I
0
a
Z
0
Positive-sequence Negative-sequence Zero-sequence
3
Here the fault occurs between line "a" and ground
through a fault impedance
f
Z . Assuming the
generator was initially at no load, the boundary
conditions at the fault are:
0
a f a
b c
V Z I
I I



Using t-he second equation in the symmetrical
components of currents we have:

0
1 2
2 2
1 1 1
1
1 0
3
1 0
a a
a
a
I I
I a a
I a a
_
_ _





, ,
,

Therefore
0 1 2
1
3
a a a a
I I I I . Thus the three symmetrical components are equal and
each is equal to one-third phase "a" current.

Phase "a" voltage in terms of symmetrical components is:

0 1 2
a a a a
V V V V + +
Using equation ++ last page, and the fact the three symmetrical components of current
are
0 1 2
1
3
a a a a
I I I I we have:

( )
0 1 2 0
a a a
V E Z Z Z I + +
Where
0
3
s n
Z Z Z + and
1 2
s
Z Z Z . Recall that
a f a
V Z I and that
0
3
a a
I I , the
above equation becomes:

( )
0 1 2 0 0
3
f a a a
Z I E Z Z Z I + +
Or:

0
1 2 0
( )
3
a
a
f
E
I
Z Z Z Z
+
+ + +

And the fault current is:

0
1 2 0
3
3
3
a
a a
f
E
I I
Z Z Z Z

+ + +

Using the symmetrical components of current in equation ++ the symmetrical
components of voltage are found, hence by transformation, the voltages during the fault.

Observing the equations
+ and ++ we notice that
the sequence networks
derived earlier may be
connected as shown here
which satisfies these two
equations.
E
a
E
b
E
c
Z
s
Z
s
Z
s
Z
n
I
a
I
b
=0
I
c
=0
I
n
+
+
+
-
- -
V
a
V
c
V
b
Z
f
E
a
+
-
V
1
a
+
-
V
2
a
+
-
V
0
a
+
-
I
1
a
Z
1
I
2
a
Z
2
I
0
a
Z
0
3Z
f
4
If the generator neutral is grounded, then 0
n
Z and
0 s
Z Z , and for a bolted fault
0
f
Z .


10.6 Line-To-Line Fault

This fault is shown in the figure to the right. Note the
neutral is not grounded. The fault is shown between
phases b and c. It is assumed the generator is initially on
no-load. The boundary conditions at the fault point are:
0
0
b c f b
b c
a
V V Z I
I I
I

+


Using 0
a
I and
c b
I I , the symmetrical components
of currents become:

0
1 2
2 2
1 1 1 0
1
1
3
1
a
a b
a b
I
I a a I
I a a I
_
_ _






, ,
,

Expanding the matrix equation we have:

( ) ( )
( )
0
1 2
2 2
0
1
3
1
3
a
a b
a b
I
I a a I x
I a a I





Note that
1 2
a a
I I from (x) above. From the equations for phase voltages (reproduced
below) we have:

0 1 2
0 2 1 2
0 1 2 2
a a a a
b a a a
c a a a
V V V V
V V a V aV
V V aV a V
+ +
+ +
+ +

Hence

( )( )
2 1 2
b c a a
f b
V V a a V V
Z I


Using the values of
1
a
V and
2
a
V from ++, and note that
2 1
a a
I I , we have:

( ) ( )
2 1 2 1
a a f b
a a E Z Z I Z I
1
+
]

And substituting for
b
I from (x) above we have:
E
a
E
b
E
c
Z
s
Z
s
Z
s
I
a
=0
I
b
I
c
+
+
+
-
- -
V
a
V
c
V
b
Z
f
5

( )
( ) ( )
1
1 2 1
2 2
3
a
a a f
I
E Z Z I Z
a a a a
+


Since
( ) ( )
2 2
3 a a a a , solving for
1
a
I we have:

1
1 2
a
a
f
E
I
Z Z Z

+ +

The phase currents are:

2 1
2 1
1 1 1 0
1
1
a
b a
c a
I
I a a I
I a a I
_ _ _

, , ,

The fault current is:

( )
2 1
1
3
b c a
a
I I a a I
j I



Using the symmetrical components of currents in equation ++, the symmetrical
components of voltage (hence phase voltages by transformation,) at the fault point are
obtained.

Note that
0
0
a
I and
2 1
a a
I I , hence from
xx above we see that the sequence network
for the line-to-line fault is as shown on the
right.



10.7 Double Line-To-Ground
Fault

This fault is not as common as the previous two,
but more likely than the symmetrical fault (all
lines bolted, or connected through a fault to the
neutral.) The double line-to-ground fault is
illustrated on the right. The fault is a short
between lines b and c, which are grounded
through fault impedance
f
Z . Again, the
generator is assumed initially at no-load. The
boundary conditions at the fault point are:
(0.1)
( )
0 1 2
0
b c f b c
a a a a
V V Z I I
I I I I
+
+ +


We know that the voltages
b
V and
c
V are given by:

E
a
+
-
V
1
a
+
-
V
2
a
+
-
I
1
a
Z
1
I
2
a
Z
2
Z
f
E
a
E
b
E
c
Z
s
Z
s
Z
s
I
a
=0
I
b
I
c
+
+
+
- - -
V
a
V
c
V
b
Z
f
Z
n
6
(0.2)
0 2 1 2
0 1 2 2
b a a a
c a a a
V V a V aV
V V aV a V
+ +
+ +


And since
b c
V V , we have:
(0.3)
1 2
a a
V V
Substituting "012" for the "abc" currents in (0.1) we have:
(0.4)
( )
( )
0 2 1 2 0 1 2 2
0 1 2
0
2
3
b f a a a a a a
f a a a
f a
V Z I a I aI I aI a I
Z I I I
Z I
+ + + + +


Using (0.4) and (0.3) in the top equation of (0.2) we have:
(0.5)
( )
0 0 2 1
0 1
3
f a a a
a a
Z I V a a V
V V
+ +


Using (++) for the symmetrical components of voltage in the equation above, and solving
for
0
a
I we have:
(0.6)
1 1
0
0
3
a a
a
f
E Z I
I
Z Z


+

Using the equation (++) in (0.3) we have:
(0.7)
1 1
2
2
a a
a
E Z I
I
Z


Now using the above two equations in the second equation of (0.1) we have (after some
algebra:)
(0.8)
( )
1
2 0
1
2 0
3
3
a
a
f
f
E
I
Z Z Z
Z
Z Z Z

+
+
+ +

Note that
a
E drives the current
1
a
I , hence the impedance seen by
a
E is:
( )
2 0
1
2 0
3
3
f
f
Z Z Z
Z
Z Z Z
+
+
+ +
. Clearly,
this is the impendence
1
Z in
series with the parallel
combination of
2
Z and
0
3
f
Z Z + . Thus the sequence
networks can be connected as
shown in the circuit on the
right. Once the sequence
currents are found from (0.8),
(0.7) and (0.6) [in that order,] the phase currents are found using the transformation
matrix A. Finally the fault current is found from:
(0.9)
0
3
f b c a
I I I I +
E
a
+
-
V
1
a
+
-
V
2
a
+
-
V
0
a
+
-
I
1
a
Z
1
I
2
a
Z
2 I
0
a Z
0
3Z
f
7
[This is seen from the equation ( )
0
1
3
a a b c
I I I I + + and knowing in this case 0
a
I .]

Example 10.5

The one-line diagram of a simple power
system is shown to the right. The neutral
of each generator is grounded through a
current-limiting reactor of 0.25/3 per unit n
a 100-MVA base. The system data
expressed in per unit on a common 100-
MVA base is tabulated below. The
generators are running on no-load at their
rated voltage and rated frequency with their
emfs in phase.

Determine the fault current for the
following faults:

(a) A balanced three-phase fault at bus
3 through fault impedance
0.1
f
Z j per unit.
(b) A single line-to-ground fault at bus 3 through fault impedance 0.1
f
Z j per unit.
(c) A line-to-line fault at bus 3 through fault impedance 0.1
f
Z j per unit.
(d) A double line-to-ground fault at bus 3 through fault impedance 0.1
f
Z j per
unit.
Item Base MVA Voltage rating
1
X
2
X
0
X
1
G
100 20 kV 0.15 0.15 0.05
2
G
100 20 kV 0.15 0.15 0.05
1
T
100 20/220 kV 0.1 0.1 0.1
2
T
100 20/220 kV 0.1 0.1 0.1
12
L
100 220 kV 0.125 0.125 0.3
13
L
100 220 kV 0.15 0.15 0.35
23
L
100 220 kV 0.25 0.25 0.7125

Preliminary calculations:

First we find the Thevenin impedance viewed from bus 3, the faulted bus. First the delta
is changed to a Y as can be seen in the figure below on the right. Now the Y-impedances
are calculated thus:
1 2
3
G
1 G
2
T
1
T
2
8
( ) ( )
( ) ( )
( )( )
1
2
3
0.125 0.15
0.0357143
0.525
0.125 0.25
0.0595238
0.525
0.15 0.25
0.0714286
0.525
S
S
S
j j
Z j
j
j j
Z j
j
j j
Z j
j





Combining parallel branches, and
adding the series branch, the positive-
sequence Thevenin impedance is:




( ) ( )
1
33
0.2857143 0.3095238
0.0714286
0.5952381
0.22
j j
Z j
j
j
+


The positive- and negative-sequence networks
are shown to the right. The only difference
between them in this case is that the source is
missing in the negative-sequence network.
Hence we have:
2 1
33 33
0.22 Z Z j
Now, the zero-sequence
network is constructed
based on the transformer
connections and is shown
in the figures to the right.
Note that the delta is
reduced to its equivalent Y
first, and then the
impedances are combined
to find the equivalent zero-
sequence circuit. The
result is the simple circuit
shown below.
Zero-sequence
j0.35


j0.25
j0.15
j0.25
j0.125
1
2
3
j0.25
1
2
3
j0.25 j0.25
j0.035714 j0.059524
j0.071428
S
j0.25
j0.35
j0.7125
j0.3
1 2
3
j0.25
j0.05
j0.05
j0.1
j0.1
j0.25
j0.25
j0.05
j0.05
j0.1
j0.1
1
2
3
S
j0.183026
j0.156881
j0.077064
E
a
+
-
j0.22
Positive-sequence Negative-sequence
j0.22
9
Now we are ready to solve the problem!
(a) Balanced three-phase fault at bus 3.

We assume the generators at no-load have a voltage of 1.0 per unit, hence the fault
current is:

( )
3
3 1
33
0 1.0
3.125 pu
0.22 0.1
=820.1 90 V
a
a
f
V
I j
Z Z j j

+ +

o


(b) Single line-to-groud fault at bus 3.

We know the sequence currents are the same and given by:

( )
0 1 2 3
3 3 3 1 2 0
33 33 33
0
3
1.0
0.22 0.22 0.35 3 0.1
0.9174 pu
a
f
V
I I I
Z Z Z Z
j j j j
j

+ + +

+ + +


And using the matrix A we have:

0 0
3 3 3
2 0
3 3
2 0
3 3
1 1 1 3 2.7523
1 0 0 pu
1 0 0
a
b
c
I I I j
I a a I
I a a I
_ _ _
_ _







, ,
, , ,


(c) Line-to-line fault at bus 3.

The zero-sequence component of current is zero, i.e.
0
3
0 I . Also from the analysis of
the line-to-line done previously we have:

( )
1 2 3
3 3
1 2
33 33
0
1
1.8519 pu
0.22 0.22 0.1
a
f
V
I I j
Z Z Z j j j

+ + + +


The fault current is:

3
2
3
2
3
1 1 1 0 0
1 1.8519 3.2075 pu
1 1.8519 3.2075
a
b
c
I
I a a j
I a a j
_
_ _ _







, , ,
,


(d) Double line-to-ground fault at bus 3.

From the equations developed earlier for the sequence currents of this fault we have:
10

( )
( )
( )
1 3
3
2 0
33 33
1
33 2 0
33 33
0
1
2.6017 pu
0.22 0.35 0.3
3
0.22
0.22 0.35 0.3
3
a
f
f
V
I j
j j j
Z Z Z
j
Z
j j j
Z Z Z

+
+
+
+
+ +
+ +


( ) ( ) ( )
1 1
2 3 33 3
3
2
33
0 1 0.22 2.6017
1.9438
0.22
a
V Z I j j
I j
Z j


( ) ( )( )
1 1
0 3 33 3
3
0
33
0 1 0.22 2.6017
0.6579
3 0.35 0.3
a
f
V Z I j j
I j
Z Z j j


+ +

And using the A matrix we have the phase currents:

3
2
3
2
3
1 1 1 0.6579 0
1 2.6017 4.058 165.93 pu
1 1.9438 4.058 14.07
a
b
c
I j
I a a j
I a a j
_
_ _ _








, , ,
,
o
o

The fault current is:
( )
3 3 3
1.9732 90
b c
I F I I +
o
.







Chapter 10: Symmetrical Components and
Unbalanced Faults, Part III


10.8 Unbalanced Fault Analysis Using Bus Impedance Matrix

We showed that the symmetrical components are independent, and for each component
there is an impedance matrix
bus
Z . Thus we can find
0
bus
Z ,
1
bus
Z and
2
bus
Z in order to deal
with each component separately. If the fault is at bus k , then we know that the k-th
diagonal element of the bus is the Thevenin impedance of the network viewed from that
bus. This then allows us to find the sequence networks, which we connect according to
the fault, and then go on to solve for the fault currents and voltages. The impedances
obtained from the
bus
Z matrices are called
0
kk
Z ,
1
kk
Z and
2
kk
Z which are connected
according to figures 10.11, 10.13, and 10.15 in the book. In the material to follow, the
subscript "a" is implied, but it is left out. It is understood that the symmetrical
components refer to phase a.


10.8.1 Single Line-To-Ground Fault Using Z
bus


Consider the single line to ground fault on line "a" through
impedance
f
Z to ground on bus k as shown in the figure. This
fault requires the zero- positive- and negative-sequence networks
for phase "a" be placed in series in addition to 3
f
Z in order to
compute the sequence currents for phase "a", thus:

( )
0 1 2
0 1 2
0
3
k
k k k
kk kk kk f
V
I I I
Z Z Z Z
= = =
+ + +

Where
0 1 2
, ,and
kk kk kk
Z Z Z are the diagonal elements in the k axis of the corresponding
impedance matrix, and
( )
0
k
V is the pre-fault voltage at bus k. The fault phase current is:

012 abc
k k
I AI =


10.8.2 Line-To-Line Fault Using Z
bus


Consider a fault between phases b and c through an impedance
f
Z
at bus k as shown in the figure. Connecting the sequence networks
for phase "a" as found earlier (positive- and negative-sequence
networks in opposition,) the symmetrical components of the fault
current are:
Z
f
a b c
Z
f
a b c
2

( )
0
1 2
1 2
0
0
k
k
k k
kk kk f
I
V
I I
Z Z Z
=
= =
+ +

Where
1
kk
Z , and
2
kk
Z are on the diagonal elements in the k-axis of the corresponding bus
impedance matrix. The fault phase current is then obtained using the A matrix as usual.


10.8.3 Double Line-To-Line Fault Using Z
bus


Here phases b and c are shorted and connected to ground through a
fault impedance
f
Z . From before, the equations for the sequence
currents of phase a at bus k are given by:

( )
( )
( )
( )
1
2 0
1
2 0
1 1
2
2
1 1
0
0
0
3
3
0
0
3
k
k
kk kk f
kk
kk kk f
k kk k
k
kk
k kk k
k
kk f
V
I
Z Z Z
Z
Z Z Z
V Z I
I
Z
V Z I
I
Z Z
=
+
+
+ +

=
+

As usual, the
0 1 2
, ,and
kk kk kk
Z Z Z are the diagonal elements in the k axis of the
corresponding impedance matrix, and
( )
0
k
V is the pre-fault voltage at bus k. The phase
currents are obtained using the A matrix, and the fault current is
( )
b c
k k k
I F I I = + .


10.8.4 Bus Voltages and Line Currents During Fault

Knowing the fault currents at bus k, the symmetrical components of the i-th bus voltage
during fault are obtained using the equation (++):

( )
( ) ( )
( )
0 0 0
1 1 1 1
2 2 2
0
0
0
i ik k
i i ik k
i ik k
V F Z I
V F V Z I
V F Z I
=
=
=

Where
( ) ( )
1
0 0
i i
V V = is the pre-fault phase voltage at bus i and
ik
Z is the ik-th element
of the proper Z matrix. The phase voltages during the fault are found using the A matrix
thus:

012 abc
i i
V AV =
The symmetrical components of fault current in line i to j are given by:
a b c
Z
f
3

( ) ( )
( ) ( )
( ) ( )
0 0
0
0
1 1
1
1
2 2
2
2
i j
ij
ij
i j
ij
ij
i j
ij
ij
V F V F
I
z
V F V F
I
z
V F V F
I
z

=

Where
0 1 2
, ,and
ij ij ij
z z z are the zero-, positive-, and negative-sequence components of the
actual line impedance between buses i and j. Using the A matrix the phase fault currents
in line i to j are:

012 abc
ij ij
I AI =

Example 10.6

Solve Example 10.5 using the bus impedance matrix. In addition, for each type of fault
determine the bus voltages and line currents during the fault.

Go over the example in detail, page 435.


10.9 Unbalanced Fault Programs

Go over the various programs for unbalanced fault analysis.