Introduction to Fault Analysis

1.0 Timing, protective systems, and fault

types
To begin, we introduce a common way of
referring to time that you may not have
heard used yet. Time is measured in cycles
where 1 cycle=1/60 of a second=0.0167
second.
Another important term is protection. A
single protective system will have at least
one circuit breaker, one relay, and
instrument
transformers
to
measure
electrical quantities on the power system.
Transmission lines generally have a
protective system one each end of the line.
Following initiation of a fault, a properly
operating circuit breaker will open the
circuit after 4-8 cycles.

A fault refers to a short circuit in a power

system. Faults can be divided into two broad
classes: temporary and permanent faults.
Temporary
faults
cause
momentary
disruption, but the fault is cleared without
protection operation. The most common
type of temporary faults are those from
lightning.
Permanent faults cause sustained disruption
if not cleared by protection. The most
common permanent fault types are
associated with one or more of the following
1.
Wind
2.
Ice loading
3.
Thermal heating and sag
4.
Various rare events such as:
a.Trees growing into line
b.
Automobile striking tower
c.Earthquakes
d.
Flooding
e.Airplanes
2

Recognizing that transmission systems are

comprised of three phases, we may observe
that a line may be faulted in a variety of
ways. The four most common ways are
1. Three phase
2. Single line to ground
3. Double line to ground
4. Line to line
Most power systems find that #2 is the most
common, comprising about 70% of all
faults. Type 4 is next, at about 15%. Type 3
is next, at about 10%.
Type #1 is least common, usually only about
5 % are 3 phase faults. However, type #1 is
a special kind of fault in that it is symmetric,
i.e., all 3 phases see the same load even
during the fault and therefore conditions
remain balanced. The fact that conditions
remain balanced is good for analysis
purposes, as well shall see, but not at all
good for the power system, as the 3 phase
fault is usually the most severe.
3

 What do we mean by severe?

2.0 Problems caused by faults
Faults cause three kinds of problems.
a.Currents: Since faults are short circuits,
they force the voltage at the fault location
to zero, so that each generator sees a low
impedance path to ground. So all internal
generator voltage appears across the
impedance between the generator and the
fault. The situation is illustrated in Fig. 1.

Fig. 1
In Fig. 1, the fault grounds the middle bus.
There are five paths to ground in this
4

figure. For the four load paths, almost all

voltage appears across the load. These
currents are small, since the load
impedance is large. For the short circuit
path, all voltage appears across the
generator and line impedances. Since these
impedances are small, currents are large. A
simpler illustration is shown in Fig. 2.

Fig. 2
In Fig. 2, we see that impedance seen by
the generator is 0.000002+j0.1, so that if
the generator voltage is 1.0, the current is
1.0/(0.000002+j0.1)=0.002-j10 pu, a large
current. Protection equipment must be able
to (a) carry the high current (for a short
time); (b) interrupt that high current.
b. Generator acceleration: From the
previous example, we see that the current
5

is 0.002-j10, resulting in a power of VI*

(V=1), which is 0.002+j10. Without the
fault the current is 1/(50+j0.09)=0.02j0.00002, resulting in a power of
0.02+j0.00002. We see that the effect of
the fault on the power is to dramatically
increase the reactive and to decrease the
real. This happens because the high
current causes an increase in voltage drop
across the reactive elements in the
generator and lines and a decrease in
voltage drop across the load impedance.
The fact that the real power consumption
of the network decreases is a problem
because the real power mechanical input
to the generator does not decrease (it
comes from the turbine, from the steam,
which is unaffected by the fault). So the
generator sees same amount of real power
in, but decreased real power out. What
does it do? Accelerate! We must remove
the faulted condition quickly, otherwise
the generator will speed up too much and
6

lose synchronism with the rest of the

network. Loss of synchronism is a very
bad thing, and we must prevent it.
c.Loss of a component: Proper action by the
protection system to eliminate the faulted
condition from the network results in loss
of a component and therefore a weakening
of the network. This can cause overloads,
undervoltages, and voltage instability.
In this course, we will study in some depth
the first two problems listed above, both of
which relate to the fault-on time period.
These two problems are generally called
1. Fault analysis
2. Transient instability
We will probably not get time to study
problem (c) about loss of a component,
although we might get close.
Closely related to both problems (a) and (b)
is a third issue that we will study

3. Protection: circuit breaker selection and

relay settings
These topics will take us up to spring break,
as observed on the course web page at
www.ee.iastate.edu/~jdm/EE457/ee457schedule.htm.

Question: why is protection closely related

to fault analysis and transient instability?
The objective of fault analysis is to
establish the requirements for the circuit
breaker, or to check that the existing circuit
breaker is adequate. Critical information
here includes the maximum current rating
and the interrupting capability of the circuit
breaker.
A key issue for transient stability is the
length of time for which the unit is
accelerating. The longer is this time, the
more likely it is that the unit will lose
synchronism. The length of time for which
the unit is accelerating is determined by the
time required for the circuit breaker to open
following the fault.
8

We will also study several other topics, all of

which relate to what is typically found in an
energy management system (EMS):
4. Automatic generation control
5. Economic dispatch with losses
6. State estimation
The amount of time we will spend on these
topics can be seen at
www.ee.iastate.edu/~jdm/EE457/ee457schedule.htm.

3.0 Transients in RL networks

A power transmission network is comprised
of elements that have primarily resistance
and inductance only (there is some
capacitance but it tends to be small
compared to the inductance).
It is
informative, therefore, to study the
characteristics of an RL circuit. Our main
goal in doing so is to see the relationship
between the DC and steady-state
components of the current after a fault.
Consider the circuit in Fig. 3.

Fig. 3
The situation we will study is analogous to
an open circuit generator that suddenly
closes into a faulted power system. The
R+jL is the Thevenin impedance seen from
the terminals of the generator looking into
the faulted power system.
Assume that the voltage source is given by
v(t ) Vm sin( t )
(1)
The parameter provides a way to control
the timing of when the switch is closed.
Using a trig identity, we can see that the
above can be written as:
v(t ) Vm (sin t cos cos t sin )
(2)

10

Lets write the voltage equation for the

circuit of Fig. 2:
di (t )
Ri (t ) L
Vm (sin t cos cos t sin )
dt
(3)

Take LaPlace transform of (3) to obtain:

s sin cos
2
2
2
2
s

(4)

RI ( s) LsI ( s ) Li (0) Vm

When the switch is just closed at t=0, we

have zero current, therefore (4) is:
s sin cos
2
2
2
2
s

RI ( s) LsI ( s ) Vm

(5)

Solving for I(s) results in:

I ( s)

Vm
L

R
/
L

s sin cos
2
2
2
2
s

(6)
Distributing the two factors through yields:
Vm
Vm
s sin
cos
L
L
I ( s)

s R / L s 2 2 s R / L s 2 2

11

(7)

We will skip some tedious steps at this

point. These steps involve:
Application of partial fraction expansion
Some algebra
Inverse LaPlace transform
These steps result in:
i (t )

where

Vm
sin t sin e Rt / L
Z

(8)

L
tan

R
1

(9)
is the power factor angle, i.e., angle by
which steady-state current lags voltage, and
Z R 2 L

(10)
is the magnitude of the Thevenin impedance.
Notice the qualitative difference between the
two terms inside the curly brackets of (8).
The first term, call it i1(t), is a sinusoidal

12

function of time, and provides that an

oscillating current is present for all time.
i1 (t )

Vm
sin t
Z

(11)
The second term, call it i2(t), is an
exponentially decreasing function of time, a
DC offset, given by:
Vm
i2 (t )
sin e Rt / L
Z

(12)

Notice that, at t=0, i2(0) is given by:

Vm
i2 (0)
sin i20
Z

(13)

so that
i2 (t ) i20 e Rt / L

(14)
So the current, i(t), is composed of i1 and i2:
i (t ) i1 (t ) i2 (t )
(15)
One important observation here is that
because the current in the inductor is
zero just before the switch closes,
then the current in the inductor must be
zero just after the switch closes.
13

The reason for this is that current cannot

change instantaneously in an inductor.
If the current could change instantaneously,
then di/dt could be infinite, making Ldi/dt
(voltage across the inductor) also infinite.
Therefore, it must be the case that i(0)=0
under all possible conditions. Since
i(0)=i1(0)-i2(0), then i1(0)=i2(0).
But (11) indicates that i1(0) depends on -.
If -=0, then i1(0)=0 and i2(0)=i20=0, and so
there will be no DC offset.
For what value of does this occur?
For a circuit where inductance is much
larger than resistance, the power factor angle
is very close to /2 (for a purely inductive
circuit, current lags voltage by 90).
This means that i1(0)=0 if switch is closed so
that , i.e., so that the voltage waveform
is at Vmsin(/2)=Vm, a positive maximum.
14

So when do we get the largest DC offset?

When i1(0) is a maximum.
This occurs when -=/2.
Given /2, if , then -/2. So we
obtain maximum DC offset when the
switched is closed so that , i.e., so that
the voltage waveform is at Vmsin()=0, a
zero
Lets take a look at some numerical data to
illustrate.
R=1 ohm
L=0.05 henry
Vm=10 volts
=2**60 radians/sec
=/2
The significance of the last bullet is that the
switch is being closed when the voltage
waveform is almost at a maximum, i.e.,
v(t ) Vm sin( t ) v(0) Vm sin( / 2) Vm
15

A plot of the voltage waveform for this

condition is given in Fig. 4.
10
8
6

Voltage (volts)

4
2
0
-2
-4
-6
-8
-10

0.02

0.04

0.06

0.08
0.1
0.12
Time (seconds)

0.14

0.16

0.18

0.2

Fig. 4
We can calculate:
tan 1 * L / R 1.5178 radians (86.96)
Z R 2 L 18.8761
2

ohms
0.0530 radians (3.04)

16

We use the following Matlab code to

compute the currents i1, i2, and i.
R=1;
L=0.05;
Vm=10;
omega=2*pi*60;
alpha=pi/2;
theta=atan(omega*L/R);
zmag=sqrt(R^2+(omega*L)^2);
t=0:.001:0.2;
i1=(Vm/zmag)*sin(omega*t+alpha-theta);
i2=-(Vm/zmag)*sin(alpha-theta)*exp(-R*t/L);
i=i1+i2;
plot(t,i1,'r:',t,i2,'g--',t,i,'b-');
legend('i1','-i2','i=i1-i2');
ylabel('current (amperes)');
xlabel('time (sec)');
grid

Figure 5 shows the result.

0.6
i1
-i2
i=i1-i2

0.4

current (amperes)

0.2
0
-0.2
-0.4
-0.6
-0.8

0.02

0.04

0.06

0.08

0.1
0.12
time (sec)

17

0.14

0.16

0.18

0.2

Fig. 5
Some observations:
The dotted red curve, i1, is the steadystate term, and oscillates for all time.
The yellow dashed curve, i2, is the DC
offset term. It is small to begin with and
goes to almost zero after about 0.1 seconds.
The blue solid curve, i, is the composite
current. It becomes the same as i1 after the
DC offset has died (after about 0.1 sec).
The DC offset is so small that it has
almost no affect on the composite current.
Now change =. In this case, the voltage
waveform is almost at a zero, i.e.,
v(t ) Vm sin( t ) v(0) Vm sin( ) 0
A plot of the voltage waveform for this
condition is given in Fig. 6.

18

.
10
8
6

Voltage (volts)

4
2
0
-2
-4
-6
-8
-10

0.02

0.04

0.06

0.08
0.1
0.12
Time (seconds)

0.14

Fig. 6
In this case, we have that
1.6238 radians (93.04)
Figure 7 shows the results.

19

0.16

0.18

0.2

0.6
i1
-i2
i=i1-i2

0.4

current (amperes)

0.2
0
-0.2
-0.4
-0.6
-0.8
-1

0.02

0.04

0.06

0.08

0.1
0.12
time (sec)

0.14

0.16

0.18

0.2

Fig. 7
Some observations:
The dotted red curve, i1, is the steadystate term, and oscillates for all time.
The yellow dashed curve, i2, is the DC
offset term and goes to almost zero after
about 0.2 seconds.
The blue solid curve, i, is the composite
current. It becomes the same as i1 after the
DC offset has died (after about 0.2
seconds).
20

 The DC offset term here is quite large. In

fact, it causes the current to reach a value at
about 0.01 second that is almost twice the
amplitude of i1.
Homework (due Wednesday):
1. Using the output from the matlab code
provided above, for =, compute the
ratio K()=|i|max()/|i1|max.
2. Repeat for the following values of :
=3, 2.5, 2, 1.5, 1, 0.5, 0.
3. Repeat parts (1) and (2) but use R=0.1.
4. Repeat parts (1) and (2) but use R=10.
The point of the above exercise is that,
depending on where on the voltage
waveform the breaker opens, the DC offset
term i2 can cause the current to be
significantly higher than the steady-state
term i1. It is should be clear from the
exercise, and from inspection of equations
(11) and (12) that an upper bound for |i|
max()/|i1|max is 2.0.
21

Gross [1, p. 360] and Glover & Sarma [2, pg.

278] analyze this situation in terms of RMS
current values, as follows.
Recall that the rms value of a periodic
function is the square root of the sum
obtained by adding the square of the rms
value of each harmonic to the square of the
DC value [3, pp. 729]. Stretching this
concept a bit, if we assume that the DC
value at some selected time t, i2(t), is
constant, we may then compute an rms value
of the composite current as

I (t ) I 12 i 2 (t )

(16)
where I is the rms value of the composite
current, and I1 is the rms value of the steadystate current.
The indicated references [1,2] then show
that the upper bound of |I|()/|I1| is
.
22

3 1.73

Thus, we see that the maximum value of rms

current is 1.73 times the rms steady-state
current I1.
We will find it very convenient to only
compute the steady-state fault current. Then
we can specify that the circuit breaker have
an interruptible rating (rms) of 1.73 times
the steady-state rms value of the fault
current.
4.1 Consideration of all three phases
We saw that the DC component depends on
where on the voltage waveform the switch is
closed. For a three-phase synchronous
generator, however, the three phase voltages
are out of phase by 120. Assuming a threephase fault shorts all three phases at exactly
the same time, then each phase sees a
different . The matlab code for studying
this situation is below, and Fig. 8 illustrates.
R=1;

23

L=0.05;
Vm=10;
omega=2*pi*60;
alpha=pi/2;
theta=atan(omega*L/R);
zmag=sqrt(R^2+(omega*L)^2);
t=0:.001:0.2;
% a-phase
i1a=(Vm/zmag)*sin(omega*t+alpha-theta);
i2a=-(Vm/zmag)*sin(alpha-theta)*exp(-R*t/L);
ia=i1a+i2a;
% b-phase
i1b=(Vm/zmag)*sin(omega*t+alpha-2.0944-theta);
i2b=-(Vm/zmag)*sin(alpha-2.0944-theta)*exp(-R*t/L);
ib=i1b+i2b;
% c-phase
i1c=(Vm/zmag)*sin(omega*t+alpha+2.0944-theta);
i2c=-(Vm/zmag)*sin(alpha+2.0944-theta)*exp(-R*t/L);
ic=i1c+i2c;
plot(t,ia,'r:',t,ib,'g--',t,ic,'b-');
legend('a-phase','b-phase','c-phase');
ylabel('current (amperes)');
xlabel('time (sec)');
grid
1
a-phase
b-phase
c-phase

0.8
0.6

current (amperes)

0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1

0.02

0.04

0.06

0.08

0.1
0.12
time (sec)

Fig. 8
24

0.14

0.16

0.18

0.2

5.0 Decaying steady-state term

In previous sections, we learned that if a
synchronous generator experiences faulted
conditions, it will respond as an RL circuit.
However, our treatment assumed that the
steady-state terms for each phase have
constant amplitude. This is not actually the
case. These terms actually have an
amplitude that decays from a maximum
value at the instant of fault initiation to a
steady-state value following some time. This
effect is due to the fact that the internal
voltage of the generator is a function of the
field current, the a-, b-, and c-phase currents,
and the flux linkages between the
corresponding windings, and the phase
currents are of course changing significantly
during the transient, and so the internal
voltage is also changing significantly.
Detailed analysis of these effects is tedious
and not worth the time it will take to do it.

25

Fig. 9 illustrates the actual response of what

we previously called i1(t) for the a-phase i.e.,
the DC offset term is not considered here.
Since we have removed the DC-offset term,
we will call this current the symmetrical rms
current.

Fig. 9
The envelope Imax(t) indicated in Fig. 9 is the
maximum symmetrical current. Division of
2 gives the rms symmetrical
Imax by
current that we previously called I1.

26

Suppose that we wish to use our R-L circuit

to compute I1(t) as a phasor, as in Fig. 10.

Fig. 10
We will get
I1

Eg
R j L

Eg
R jX

(17)

If we are interested only in current

magnitude, then R+jXjX and so
I1

Eg
X

(18)
But of course, (18) only gives us a single
value of the current magnitude, and clearly
the current magnitude decreases with time
during the first few cycles.

27

We could always compute the exact

transient as shown in Fig. 9. However, in
order to enable simpler analysis, we will
define three different generator reactances to
use in approximating the rms symmetrical
current. These are:
Xd: subtransient reactance, used to
approximate current from t=0+ to t=2
cycles
Xd: transient reactance, used to
approximate current from t=2 cycles to
t30 cycles.
Xd: steady-state reactance: used to
approximate current during the steadystate, which is generally after about 30
cycles.
Therefore we obtain three different currents
corresponding to these three different
reactances. They are:
I: subtransient current
I: transient current
28

 I: steady-state current
The situation is illustrated in Fig. 11-a and b.

Fig. 11
In Fig. 11-a, we have discretized the current
into three intervals corresponding to I, I,
and I.
Correspondingly, in Fig. 11-b, we have
discretized the generator reactance into three
intervals corresponding to X, X, and X.
If the generator is unloaded when the fault
occurs, with internal voltage of Eg, then the
29

three different fault currents may be

computed by:
Eg

X d''

I
I

(19)

Eg
X d''

(19)

Eg
Xd

(19)
The above analysis applies to a smooth-rotor
machine. It is more complex for a salient
rotor machine, but smooth rotor analysis
gives good approximations for a salient pole
machine.

30

Additional HW assignment for Wednesday:

Two generators are connected in parallel to
the low-voltage side of a three-phase -Y
transformer, as shown in the figure below.
Generator 1 is rated 50,000 kVA, 13.8kV.
Generator 2 is rated 25,000kVA, 13.8kV.
Ech generator has a subtransient reactance
of 25% on its own base. The transformer is
rated 75,000kVA, 13.8/69Y kV, with a
reactance of 10%. Before the fault occurs,
the voltage on the high-voltage side of the
transformer is 66kV. The transformer is
unloaded and there is no circulating current
between the generators. Find the
subtransient current in each generator when
a three-phase short circuit occurs on the
high-voltage side of the transformer.

31

Hint: The circuit to analyze should appear as

below.

32

1[] C. Gross, Power System Analysis, second edition, Wiley, 1986.

2[] J. Glover and M. Sarma, Power system analysis and design, PWS Publishers,

Boston, 1987.
3[] J. Nilsson, Electric Circuits, second edition, Addison Wesley, 1986.