Fault Analysis Using Z-bus

1.0 Introduction

The previous notes on Z-bus resulted in

some useful knowledge: diagonal element
Zkk of the Z-bus is the Thevenin impedance
seen looking into the network from bus k.

Combining this knowledge with equation (4)

from notes called “Symmetrical Faults 2”,
which was
V
I  
(1)
f
f
f Z Z Thev

enables us to efficiently obtain the fault

current for any bus in the network.

This is easy if we have Z-bus and Vf.

However, one other thing that we will

frequently need is the line currents, because
the circuit breakers are going to be in series
with the lines, not with the faults.
2.0 Fault calculations using Zbus
1
Recall that
V  ZI
(2)
Since (2) represents a set of linear equations,
superposition holds, and we may write:
V  Z  I
(3)
This says that the change in voltage at all
buses ΔV may be computed if the change in
injections at all buses ΔI are known. We can
write eq. (3) in expanded form as:
 V1   Z11  Z1k  Z1N   I1 
            
  
 Vk    Z k 1  Z kk
  
 Z kN   I k 
  (4)
           
VN   Z N 1  Z Nk  Z NN  I N 

Now consider a fault at bus k, where the pre-

fault voltage at bus k is Vf. Let the fault
current be I’’f, and assume that the fault
impedance Zf=0 (this is typically worst-case
scenario).
Since a fault is a short circuit, then ΔVk=-Vf.
Also, since the fault current is out of bus k,
then ΔIk=- I’’f. Substituting these into eq. (4)
results in

2
  V1   Z11  Z1k  Z1N   0 
          
  
   V f    Z k1
row k
 Z kk  Z kN   I f  (5)
    
         
VN  Z N1  Z Nk  Z NN   0 
Note that the right-hand-side results in, for
each row j, only the Zjk being multiplied by
a non-zero current. Therefore:
 V1    Z1k I f 
    
   
rowk   V f     Z kk I f  (6)
   
    
VN   Z Nk I f 
We observe from row k that:
 V f   Z kk I f
(7)
Solving (7) for I’’f results in
Vf
I f 
(8)Z kk

Notice that eq. (8) is consistent with eq. (1)

when Zf=0.

Now substitute eq. (8) into eq. (6) to get:

3
  V1     Z1k / Z kk V f 
    
   
rowk   V f     V f  (9)
   
    
VN    Z Nk / Z kk V f 
 

Now eq. (9) provides the change in the bus

voltages due to the fault. Change from what?

It is the change from the voltage without the

fault, i.e., it is the pre-fault voltage.

Consider any bus, let’s say bus j, with a pre-

fault voltage of Vj. Then we can compute the
bus j voltage under the faulted condition as
V  V  V
jf j
(10) j

From eq. (9), we know that

Z
V  
Z
V j (11) jk

kk
f

Substitution of (11) into (10) results in

Z
V V jf
Z
V j (12)
jk

kk
f

Now eq. (12) is useful for computing fault

currents in the circuits. Consider Fig. 1.
4
 Fig. 1

We can use eq. (12) to write down the

voltages under the faulted condition for
buses i and j, as
Z
V V 
if i
Z
V ik

kk
(13) f

Z
V V 
jf j
Z
V
kk
(14)
jk
f

Now we can compute the subtransient

current flowing from bus i to bus j under the
fault condition as
V V
I   ij
Z (15)if

b
jf

Substituting eqs. (13) and (14) into (15)

results in
Z ik Z jk
Vi  V f V j  Vf
Z kk Z kk
I ij 

Vi  V j
Zb
Z ik  Z jk
(16)
 V f
Zb Z b Z kk

We can use eq. (16) to get the fault current

in the circuits. These values provide us with

5
the appropriate information for selecting the
circuit breakers in the lines.

3.0 Some important comments

 Zbus should be developed using
subtransient reactances in generator/motor
models.
 Because fault currents are typically
much larger than load currents, it may be
assumed that there are no loads.
o All pre-faults currents are 0.
o All buses have voltage (pre-fault) equal
to Vf.
o Equation (16) becomes:
Z Z
I   V
ij f
ik

Z Z b
(17)
kk
jk

 From (16) and (17), we see that only the

kth column of the Z-bus is required to
analyze a fault at bus k.

The last observation can be utilized in an

effective fashion when performing fault
analysis. Let’s assume that we want to

6
compute the short circuit currents for a fault
at only one bus k. So we just want to get the
kth column of Z-bus, but we do not need the
entire Z-bus.

There is an efficient way to get the k th

column of Z-bus. Let’s study it.

Consider that the Z-bus and the Y-bus are

inverses of each other, i.e.,
Z Y
1
(18)
This means that their product gives the
identity matrix.
YZ  I
(19)
where I is given by a matrix of zeros except
the diagonal which contains all ones, i.e.,
1 0  0
0 1  0
I 


  

(20)
0 0  1

Our approach will depend on two ideas:

1. Column of Z: We can just consider a
single column of Z, instead of the entire
matrix. Call it Zk; it is a column vector.
The right hand side of (19) will just be a
7
 column of I. Call it Ik. It will also be just
a vector and will contain zeros in every
row except for row k. The resulting
relation is:
YZ  I
k k (21)
2. LU-Decomposition: If you have taken
EE 456 or a linear algebra course in
math, then you are familiar with LU
decomposition. LU decomposition
provides a way to solve for the vector x
in the matrix relation
Ax  b
(22)
where A is a n×n square matrix, x is an
unknown n×1 vector, and b is an known
n×1 vector. The advantage to LU
decomposition is it does not require
inverting the matrix A. The basis of LU-
decomposition is that we may factor A
into a matrix product LU, i.e.,
A  LU
(23)
where L is a lower diagonal n×n matrix
of the form

8
 l11 0 0
L  l21 l22 0  (24)
l31 l32 l33 

and U is an upper diagonal n×n matrix of

the form
1 u12 u13 
U  0 1 u23 
0 0 1 
(25)
Substitution of eq. (23) in (22) yields
LU x  b
(26)
Defining
Uxw
(27)
provides that eq. (25) becomes:
Lw  b
(28)
If we have L and U, then (28) is easily
solved for w (without inverting L) using
forward substitution, and then (27) can
be easily solved for x (without inverting
U) using backwards substitution. More
details on LU decomposition may be
found in the notes called
“LUDecomposition.doc.”

Now observe that eq. (21) and (22) are in the

same form. Therefore we want to solve the

9
following equations in the order they are
given:
Lw  I (29) k

UZ  w (30) k

Factorization of a matrix Y can be done efficiently and easily using the matlab
command:
[L,U]=lu(Y)
Then it is easy to find w by hand using forward substitution from:
Lw  I k
And then it is easy to find Z2 by hand using backwards substitution from:
UZ2  w
Alternatively, the manual steps of LU decomposition can be performed per the notes
from “LU Decomposition.”

Homework #3: Due Tuesday, Feb 3.

1. Work problem 9.8 in book.

2. Consider the 4-bus system shown
below. Both machines have subtransient
reactances of 0.20 pu (you can combine
the machine subtransient reactance with
the transformer impedance to get a
single reactance connecting the machine
internal voltage with the network).

10
 a. Construct the Y-bus for this network (should be
a 4×4 matrix).
b. Consider that there is a three-phase
(symmetrical) fault at bus 2.
i. Use LU decomposition to obtain the 2nd
column of the Z-bus.
ii. Compute the subtransient fault current.
iii. Use eq. (12) to find the voltages during the
fault.
iv. Use eq. (17) to find the subtransient currents
in lines 3-2, 1-2, and 4-2.
3. See notes called “Examples.”
4. See notes called “Examples.”

11