Faults in power systems

Prof. Alberto Berizzi Dipartimento di Energia

Ph. 02 2399 3728
Email: [email protected]
Fault Analysis

• The main cause of electric power system faults is insulation

breakdown
• This breakdown can be due to a variety of different factors
– lightning
– wires blowing together in the wind
– animals or plants coming in contact with the wires
– salt spray or pollution on insulators
– Also, mechanical reasons can drive to faults like works on the roads
breaking cables and cause contacts, etc.
– Faults can be divided into
– Parallel faults (short-circuits)
– Series faults (interruptions, malfunctioning of breakers, conductor
interruptions)
– From the system point of view, also contingencies like generator
failures or bus bar splitting are important

© Alberto Berizzi - Dipartimento di Energia

What is the problem? 3

• What happens when there is a fault in the system?

• https://nos.nl/video/2443017‐explosies‐en‐rokende‐kabels‐bij‐verdeelstation‐in‐
dronten

• Why is it interesting for engineers?

• What are consequences of a fault?
• How could we prevent it?
• How can we manage it?

© Alberto Berizzi - Dipartimento di Energia

 4
Short-circuit

• It is the incidental or intentional contact, with zero or very low

impedance, between two or more points of a circuit that normally are
at different voltages
• Short-circuit current is the overcurrent resulting from a short-circuit
due to a fault or due to an incorrect connection in an electric circuit

© Alberto Berizzi - Dipartimento di Energia

Fault Types

• There are two main types of faults

– symmetric faults: system remains balanced; these faults are
relatively rare, but are the easiest to analyze so we will consider
them first
– unsymmetric faults: system is no longer balanced; very common,
but more difficult to analyze than symmetrical faults
• Most common type of fault on a three-phase system by far is
 the single line-to-ground (SLG), followed by
 the line-to-line faults (LL),
 double line-to-ground (DLG) faults, and
 balanced three phase faults
• On very high voltage lines, faults are practically always SLG, due to
large conductor spacing

© Alberto Berizzi - Dipartimento di Energia

 Why fault Analysis?

• Goal of fault analysis is to determine the magnitude of currents

during the fault
– need to determine the maximum current to ensure that devices
and machines can survive the fault (mechanical point of view)
– need to determine the maximum current to be broken, to correctly
size CBs or fuses
– need to determine the maximum ground current, to size ground
plants (rods, wires, etc.)
– Conductor sizing
– Relais setting
– Insulation coordination for overvoltages
• Faults are dynamic processes; however, for some applications,
steady-state analysis can be adopted, being on the safe side
 Moreover, performing dynamic studies often require data difficult
to find

© Alberto Berizzi - Dipartimento di Energia

RL Circuit Analysis

To understand fault analysis, we need to review the behavior of an RL

circuit

v(t ) 
2 V sin(t   )

Before the switch is closed, i(t) = 0

When the switch is closed at t=0, the current will have two
components:
1) a steady-state value
2) a transient value

© Alberto Berizzi - Dipartimento di Energia

 Time varying current

Fault current components [kA]

10

-5 it
ir
-10
0 0.01 0.02 0.03 0.04 0.05 0.06
Time [s]
Fault current[kA]
20

10

-10
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time [s]

© Alberto Berizzi - Dipartimento di Energia

Generator Modeling

The time varying reactance is typically approximated

using three different values, each valid for a different
time period:
X"d  direct-axis subtransient reactance
'
Xd  direct-axis transient reactance
Xd  direct-axis synchronous reactance

© Alberto Berizzi - Dipartimento di Energia

 Generator Modeling, cont’d

• For a balanced three-phase fault at the generator terminals, the ac

fault current is:

 1  1 1  
t  1 1  
t 
iac  2 Ed   '  e Td'
  ''  '  e Td''
 sin  t   
 X d  X d X d  X
 d X d

 
where
Td'' is the direct axis subtransient time constant (  0.035 s)
Td' is the direct axis transient time constant (  1 s)

Ed
I ac (0)   I" is the subtransient fault current
X d''

© Alberto Berizzi - Dipartimento di Energia

Generator Short Circuit Currents

© Alberto Berizzi - Dipartimento di Energia

Generator Short Circuit Currents

© Alberto Berizzi - Dipartimento di Energia

Typical assumptions – on the safe side! 13

• For the sake of simplicity, the following assumptions can be made:

 Transmission lines are modelled by their series reactances;
resistances and shunt admittances are neglected
• In distribution systems, resistances might be not negligible
 Transformers are represented by their leakage reactances;
winding resistances and magnetizing branches are neglected
 Generators are represented by constant voltage sources behind
their subtransient reactances; armature resistances, saliency and
saturation are neglected
 Induction motors are considered by means of their locked rotor
reactances; they can also be neglected if they are less than about
40 kW
 All non-rotating loads are neglected

 These assumptions lead to a 0.5%-4% error

© Alberto Berizzi - Dipartimento di Energia

Typical assumptions, cont’d 14

• Pre-fault voltages are assumed equal to c p.u.

 with c=1.1 when maximum current is searched
 with c=0.9 in case of minimum short-circuit current computation
• Tap-changers are considered at the neutral tap setting

• Therefore, the short-circuit is studied as it was a steady-state

phenomenon (more precisely, a “subtransient steady-state”) although
it is not a steady-state at all

© Alberto Berizzi - Dipartimento di Energia

Fault Analysis Solution Techniques

• Circuit models used during the fault allow the network to be

represented as a linear circuit
• There are two main methods for solving for fault currents:
1. Direct method: use pre-fault conditions to solve for the
internal machine pre-fault voltages; then apply fault and solve
directly
2. Superposition and Thévenin equivalent: the fault is
represented by two opposing voltage sources; solve the two
resulting systems and superimpose
– first system voltages just represent the pre-fault operating
point
– second system only has a single voltage source

© Alberto Berizzi - Dipartimento di Energia

 Network Fault Example

For the following network, assume a fault at the terminal of the

generator; all data is per unit (Sb=100 MVA) except for the
transmission line reactance

Vn=138 kV

Xl=19.5 
xt1=0.1 xt2=0.1 Motor
Generator Sn=100 MVA
Vg=1.05 x’’M=0.2
Sn=100 MVA xM=1
x’’ G=0.15
xG=1
supplying
100 MVA@pf 0.95 lagging

19.5
xl  2
 0.1 pu
138
100

© Alberto Berizzi - Dipartimento di Energia

Direct method 17

• We start from the steady-state pre-fault conditions, where generators

and synchronous motors are modelled by an emf EG behind a
synchronous reactance XG
• We want to compute the worst conditions, i.e., the highest short-circuit
currents. This corresponds to the subtransient conditions, modelled by
an emf E”G behind a subtransient reactance X”G
• This change in model requires initial computations to compute IG and
IM

© Alberto Berizzi - Dipartimento di Energia

Initial conditions 18

• First, we need to compute the initial values for the currents: generator

For the generator

VT  1.05, SG  1.018.2
SG  1.0  18.2 
IG     0.904  j 0.297
VT  1.05 
EG  VT  jX G I G  1.349  j 0.904 [not needed]
© Alberto Berizzi - Dipartimento di Energia
Initial conditions, cont’d 19

• First, we need to compute the initial values for the currents: motor

For the motor

I M  I G  0.904  j 0.297
VM  VT  jX ext I G  1.05  j 0.3  0.904  j 0.297  
 0.961  j 0.271
EM  VM  jX M I G  0.664  j1.176 [not needed]
© Alberto Berizzi - Dipartimento di Energia
 Subtransient network: generator

• Now, considering the subtransient parameters, we need to compute

the subtransient emfs that will take place when the fault occurs
• To do so, we need to perform again the same computations on the
subtransient network, keeping the same VT, IG and IM (as currents
cannot change instantaneously) but introducing subtransient
reactances and e.m.f.s

I G  0.904  j 0.297
EG"  VT  jX G" I G  1.05  j 0.136  1.103  7.06
© Alberto Berizzi - Dipartimento di Energia
Subtransient network: motor

I G  0.904  j 0.297
" "
EM  VM  jX M I G  0.901- j 0.452  1.008  26.6
EG" and EM
"
are kept constant from this point on

© Alberto Berizzi - Dipartimento di Energia

Short-circuit current computation

When the short-circuit takes place, we can solve the two

independent linear circuits:
EG"
I Gf   0.904 - j 7.297  7.353  82.9
j 0.15
"
EM
I Mf   -0.904-1.803i=2.016  116.6
j 0.5
I f  I Gf  I Mf  7.353  82.9  2.016  116.6   j 9.1
© Alberto Berizzi - Dipartimento di Energia
 Superposition approach 23

• According to the direct method, if pre-fault conditions change, all

computations (currents and e.m.f.) are to be repeated from
scratch
• Another approach is possible: we model the fault as follows

• The fault is represented by two equal and opposite voltage sources,

each with a magnitude equal to the pre-fault voltage in the faulted
point
© Alberto Berizzi - Dipartimento di Energia
Superposition approach cont’d 24

• Then, the circuit, linear, is decomposed as follows

• The first is the pre-fault state; the second is called Thévenin state

• In the pre-fault state,

computed by a PF, the fault
current is zero, and we need
not care about E”s

• The Thévenin network is

made by the dead network
and one voltage source set at
the pre-fault voltage: note that
one pre-fault voltage, only at
the fault point, is necessary
© Alberto Berizzi - Dipartimento di Energia
Superposition approach cont’d 25

• In the Thévenin state, the solution is quite easy

© Alberto Berizzi - Dipartimento di Energia

 Two Bus Superposition Solution

Before the fault we had Ef  1.050,

I G   I M  0.952  18.2 and I M  0.95218.2
By superposition we have
I f  0  I fTh   j 9.1
I Gf  I G  I GTh  0.904  j 0.297  j 7  0.904  j 7.297
I Mf  I M  I MTh  0.904  j 0.297  j 2.1  0.904  j1.803

This matches what we calculated earlier

The superposition of course holds also for computing
voltages
© Alberto Berizzi - Dipartimento di Energia
General comments 27

• When load changes, if we use superposition, only the pre-fault state

(voltage at the fault point) needs to be re-computed
• Often, the pre-fault network is considered at no load, i.e., all bus
voltages are at the same value, equal to the pre-fault value in the
Thévenin network: in this case,
 superposition on currents is trivial
 superposition on voltages is also very simple
 from the pratical point of view, only the Thévenin solution is really
necessary
• If all we need to compute is the fault current, it can be computed
from the pre-fault value of voltage at the faulted bus and the
equivalent Thévenin impedance
• When also currents in every branch and all bus voltages are to be
computed, it is necessary to make use of superposition of both
 Pre-fault conditions and
 Thévenin conditions
© Alberto Berizzi - Dipartimento di Energia
Symmetric three-phase fault in large 28
systems

• We can adopt the

single-phase pu
equivalent circuit,
and the pre-fault
system is given
by the passive
grid and by the
bus current
injections

© Alberto Berizzi - Dipartimento di Energia

 29
Definition of bus p

• The generic bus p is not necessarily a power system bus

• V=0 means E=0

© Alberto Berizzi - Dipartimento di Energia
Procedure for short-circuit computations 30

• The procedure is the same

as in the example
• For a short-circuit at bus p,
the solid short-circuit is
modelled as the sum of
two opposite ideal voltage
sources, equal to the pre-
fault voltage at the same
bus
• Then, superposition is
considered:
 Pre-fault system
conditions
 Thévenin equivalent
system

© Alberto Berizzi - Dipartimento di Energia

 Thévenin equivalent 31

• In the Thévenin equivalent,

 the network is dead (passive) and
 all synchronous machines are modelled by means of their
subtransient impedances, included into the dead network
0
 there is only one voltage source, equal to the pre-fault voltage V p  V f
 the current exiting the faulted bus is the short-circuit current

© Alberto Berizzi - Dipartimento di Energia

 [Z] and Thévenin equivalent 32

• Let’s recall the Thévenin equivalent:

[V 0 ]  [ Z ][ I 0 ]
is the pre-fault state. Now the fault results
in a change in the injected current I p
The vector of current injections changes

 
[V ]  [ Z ] [ I 0 ]  [I ]  [V 0 ]  [ Z ][I ]

• [V] is the voltage  V1   Z11  Z1 p  Z1n   0 

change caused by the     
    Z 21 Z2 p Z 2n 
  
change in current at  Vk      I  I f 
  
0 
 p 
bus p (all complex  V p  V p   Z p1 Z pp Z pn   0 
numbers)     
      

 Vn   Z n1  Z np  Z nn   0 
  
© Alberto Berizzi - Dipartimento di Energia
 Thévenin equivalent for the faulted bus p 33

𝛥𝑉 𝑍
⋮ ⋮
𝛥𝑉 𝑍
𝛥𝑉 𝛥𝐼 ⇒ 𝑉 𝑉 𝛥𝑉
𝑍
⋮ ⋮
𝛥𝑉 𝑍

𝑉 𝑉 𝑍 𝛥𝐼 𝑉 𝑉 0

𝑉 𝑉 𝑍 𝐼

𝑉
⇒𝐼
𝑍

© Alberto Berizzi - Dipartimento di Energia

 Computation of the other variables 34

𝑍
𝑉 𝑉
𝑍 𝑉 𝑍 𝐼 𝑍
𝑉 𝑉 𝛥𝑉 𝑉
⋮ ⋮ ⋮
⋮ ⋮ ⋮ ⋮ 𝑍
𝑉 𝑉 𝛥𝑉 𝑉 𝑍 𝑉 𝑍 𝐼 𝑉 𝑉
𝑉 𝛥𝑉 𝐼 𝑍
𝑉 𝑉 𝑍 𝑉 𝑍 𝐼 0
⋮ ⋮ 0
⋮ ⋮ ⋮ ⋮
𝑉 𝛥𝑉 ⋮
𝑉 𝑉 𝑍 𝑉 𝑍 𝐼 𝑍
𝑉 𝑉
𝑍

• The post-fault voltage at the generic unfaulted bus is

given by the superposition of voltages in the pre-fault
state and voltages in the Thévenin conditions
• Note that only the p-th column of [Z] is needed, not the
complete [Z]

© Alberto Berizzi - Dipartimento di Energia

 A common assumption 35

• Often, we assume that the system is at no load that before the fault
• In this case, before the fault, there are no power flows, no currents
and every bus has the same voltage, Vf
• Then,

© Alberto Berizzi - Dipartimento di Energia

Computing branch currents 36

• Once all post-fault voltages are known, it is easy to compute all

subtransient branch currents (contributions to the fault), assuming one
branch for each connection ij
• For example, we can compute the current between generic busses i
and j

Vi  V j Vi0  V j0 V p0  Z ip  Z jp 
I ij"     
zij zij zij  Z pp 


• If the pre-fault system is assumed at no load, the first term is zero

• Only three elements of the p-th column of [Z] are needed, those
relevant to positions i, j and p (sparse inverse)
• To compute all currents, in general, only elements of the p-th column
are needed

© Alberto Berizzi - Dipartimento di Energia

Physical meaning for a pair of busses 37

• Before the fault, the Thévenin

equivalent is open at bus p
• After the fault, current and
voltage computations can be
performed according to the
circuit theory
• The results of course are
the same
• If the fault is not solid but with a
fault impedance zf, then the only
difference is that

𝑉 𝑉
𝐼
𝑍 𝑍 𝑍 𝑧 𝑍 𝑧

© Alberto Berizzi - Dipartimento di Energia

 Fault along a line 38

• This fault can be studied without

modifying [Z]
• Let’s assume we want to study a
fault in line k-p, which has
impedance zb included in [Z]
• First, we need to remove branch
k-p, by adding (-zb) to the circuit
• Then, a new bus q is introduced,
connected to both bus k and bus
p by means of zb1 and zb2; of
course, zb1+zb2=zb
• It is worth noticing that looking at
the Thévenin equivalent, it is
clear that only elements Zkk, Zkp,
Zpp of the original [Z] are needed
• Finally, short-circuit is applied to
bus q
© Alberto Berizzi - Dipartimento di Energia
 39
Conclusions on the superposition method

• Steady-state conditions are assumed (no dynamics!)

• To compute If, the pre-fault voltage at the faulted bus and the
Thévenin network are sufficient (the branch representing the
fault is not present in the pre-fault conditions)
• To compute any other voltage or current, it is necessary to
superimpose :
 Pre-fault and Thévenin currents branch by branch
 Pre-fault and Thévenin voltages bus by bus

© Alberto Berizzi - Dipartimento di Energia

Three Gen System Fault Example

For simplicity, assume the system is unloaded

before the fault with
Eg1  Eg 2  Eg 3  1.050
Hence all the prefault currents are zero
© Alberto Berizzi - Dipartimento di Energia
 Three Gen Example, cont’d

 15 10 0
Y  j  10 20 5 
 
 0 5 9 

1
 15 10 0
Z  j  10 20 5 
 
 0 5 9 
 0.1088 0.0632 0.0351
 j 0.0632 0.0947 0.0526 
 
 0.0351 0.0526 0.1409 
© Alberto Berizzi - Dipartimento di Energia
 Three Gen Example, cont’d

1.05
For a fault at bus 1 we get I f    j 9.6   I1
j 0.1088
 0.1088 0.0632 0.0351  j 9.6   1.05180 
VTh  j 0.0632 0.0947 0.0526   0    0.60180 
    
 0.0351 0.0526 0.1409   0   0.337180
1.050  1.05180   00 
V post  1.050   0.60180   0.4440
     
1.050 0.337180  0.7130 

© Alberto Berizzi - Dipartimento di Energia