SHORT CIRCUIT ANALYSIS IN UNBALANCED DISTRIBUTION NETWORKS

R. Ebrahimi, S. Jamali, A. Gholami and A.Babaei

Iran University Science and Technology, Iran

ABSTRACT

In this paper a short circuit analysis algorithm for radial three-phase distribution networks, based on two relationship
matrices method, is presented. Two relationship matrices, the bus-current-injection-to-branch current matrix and the
bus-current-to-bus-voltage matrix are used to represent the special topological characteristics of distribution networks.
These two matrices can be accomplished by simple building algorithms and are easily implemented. The proposed short
circuit analysis method is developed from these two matrices and can be used to analyze the various types of
unsymmetrical faults only by one model. This model is included from four impedances that their amounts can be set
from zero to extreme. Therefore, each kind of unsymmetrical faults is modelled by adjusting these impedances suitably.
Since the proposed method does not use the traditional admittance matrix and only uses one model for fault
calculations, therefore, the proposed method can achieve the advantages of reducing computation time, improving
accuracy and efficiency, with lower memory requirements especially. Simulation results such as phase current and post-
fault voltage profiles obtained using the proposed technique and IEEE 34-bus network are presented and discussed.

Keywords: Short circuit analysis method, Unsymmetrical fault, Unbalanced Distribution networks, Four impedances
model.

1 INTRODUCTION remained popular for several decades. However, since

distribution feeder circuit configurations are basically
Even the best designed electric systems occasionally unbalanced, this method is not well suited [2].
experience short circuits resulting in abnormally high
currents. Overcurrent protective devices, such as circuit Distribution system configurations are basically
breakers and fuses, should isolate faults at a given unbalanced; therefore, a short-circuit-analysis method
location safely with minimal circuit and equipment based on the actual three-phase representation will be
damage and minimal disruption of the plant’s operation. more accurate. From that aspect, several short-circuit-
Other parts of the system, such as cables, busways, and analysis methods based on phase co-ordinates, hybrid
disconnecting switches, shall be able to withstand the compensation method and three-phase impedance
mechanical and thermal stresses resulting from matrix have been proposed[3]-[5].
maximum flow of short-circuit current through them.
The magnitudes of short-circuit currents are usually In this paper a short circuit analysis algorithm for radial
estimated by calculation, and equipment is selected three-phase distribution networks is presented. The
using the calculation results. proposed short circuit analysis method is developed
from these two matrices the bus-current-injection-to-
Also as the distribution system becomes more heavily branch current matrix and the bus-current-to-bus-
loaded and the need and ability to reconfigure the voltage matrix that can be used to analyze the various
system for service restoration, load balancing and loss types of unsymmetrical faults only by one model. This
reduction grows; the network configuration will be model included from four impedances that their amount
changed more frequently. With each change protection can be set properly. Therefore, each kind of
device settings in the system may need updating. unsymmetrical faults is modelled by adjusting these
Therefore, there is a need for fast and more accurate impedances suitably. Simulation results such as
short circuit calculations. calculation phase currents, fault path and post-fault
voltage profiles obtained using the proposed technique
The computation of short-circuit currents for and IEEE 34-bus network are presented and discussed.
unbalanced fault in normally balanced three-phase
system has traditionally been accomplished by the
application of symmetrical components [1]. The main 2 UNBALANCED THREE-PHASE LINE MODEL
computational advantage of the conventional
symmetrical component method is that the three The determination of impedance values for three phase
sequence matrices are treated separately. For this lines is a critical step before the short circuit analysis of
reason, fault analysis based on this method has a distribution feeder can began. Figure 1 shows a three

UPEC 2007 - 942

Authorized licensed use limited to: University of Moratuwa. Downloaded on September 07,2021 at 05:46:12 UTC from IEEE Xplore. Restrictions apply.
phase line model between bus i and j. Step 1: For a distribution system with n buses (total
available phase-buses), form BIBC matrix in
dimension n × n . So the equivalent system and
substation transformer impedances consider in BIBC.

Step 2: Number the phase-lines between buses 1 to n

from root bus, respectively.

Step 3: If a line section (Bk) is located between bus i and

bus j, copy the column of the k-th bus of BIBC matrix
to the column of the m-th bus. k is the number of
Figure 1 Three phase line model. connected line to bus i-th and m is the number of line
that is between two buses.
A 4 × 4 matrix, which takes into account the self and
mutual coupling effects of the unbalanced three-phase Step 4: For keeping upper triangular property of BIBC
line section, can be expressed as equation (1). matrix fill a +1 to the position of the m-th row and
⎡ Zˆ aa Zˆ ab Zˆ ac Zˆ an ⎤ column.
⎢ˆ ⎥
[ ]
ˆ
Z abcn = ⎢ Z ba Zˆ bb Zˆ bc Zˆ bn ⎥
⎢ Zˆ Zˆ cb Zˆ cc Zˆ cn ⎥
(1) Step 5: Repeat steps 2 to 4 until all lines sections are
include in the BIBC matrix. Another matrix is BCBV
⎢ ca
⎥
ˆ ˆ ˆ ˆ that associated bus voltages to branch currents and can
⎣⎢ Z na Z nb Z nc Z nn ⎦⎥
be written as (5).
This matrix is reduced to a 3× 3 impedance matrix using
[V0 ] − [V ] = [BCBV ][B] (5)
the Kron reduction technique, so the effects of the
neutral or ground wire are still included in this model as Where V and V0 are the vectors of bus voltages and
shown in (2). no-load bus voltages, respectively. Equation (5) can be
rewritten as (6).
⎡ Z aa Z ab Z ac ⎤
[∆V ] = [BCBV ][B] (6)
[Z abc ] = ⎢⎢ Z ba Z bb Z bc ⎥⎥ (2)
Where [∆V ] = [V0 ] − [V ] .
⎢⎣ Z ca Z cb Z cc ⎥⎦
As BIBC matrix, the building algorithm for BCBV
For single-phase and two-phase line sections, the matrix in three phase distribution networks can be
3 × 3 impedance matrix will be characterized by a row developed as follows:
and column of zeros occupying the position of the
missing phase or phases. The relationship between bus Step 1: For a distribution system with n buses (total
voltages and branch currents in figure 1 can be available phase-buses) form BCBV in dimension n × n .
expressed as equation (3).
⎡Va ⎤ ⎡V A ⎤ ⎡ Z aa Z ab Z ac ⎤ ⎡ I a ⎤ Step 2: Fill first to third rows by the total equivalent
⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ system and the substation transformer impedances.
⎢Vb ⎥ = ⎢VB ⎥ − ⎢ Z ba Z bb Z bc ⎥ ⎢ I b ⎥ (3)
⎢⎣Vc ⎥⎦ ⎢⎣VC ⎥⎦ ⎢⎣ Z ca Z cb Z cc ⎥⎦ ⎢⎣ I c ⎥⎦ Step 3: If a line section (Bk) is located between bus i and
bus j, copy the row of the k-th bus of BIBC matrix to
3 RELATIONSHIP MATRICES the row of the m-th bus. k is the number of connected
line to bus i-th and m is the number of line that is
According to radial structures of distribution feeders, between two buses.
two relationship matrices, the bus-current-injection-to-
branch-current matrix (BIBC) and branch-current-to- Step 4: From equation (3) fill self impedance to the
bus-voltage matrix (BCBV) were derived and used to position of the m-th row and column. If mutual
observe the network relationship [6]. The detailed impedances be non-zero, put them in horizontal
derivation of these two matrices can be found in [7]. In direction.
this paper only the building algorithms of these two
matrices are shown. The relationship of bus current Step 5: Repeat steps 2 to 4 until all lines sections are
injections and branch currents can be expressed as (4). include in the BCBV matrix. It can also be seen that the
[ B ] = [ BIBC ][ I ] (4) building algorithms of these matrices are similar and
Where B and I are the vectors of branch currents and easy. According to definitions, BIBC matrix can be
bus current injections, respectively. The property of the used for calculation branch currents, which are
constant BIBC matrix is an upper triangular matrix that generated by the fault bus currents. Also, the
has non-zero entries of +1 only. The building algorithm corresponding variations of the bus voltages, which are
for BIBC matrix in three phase distribution networks generated by the fault branch currents, can be found by
can be developed as follows: BCBV matrix directly.

UPEC 2007 - 943

Authorized licensed use limited to: University of Moratuwa. Downloaded on September 07,2021 at 05:46:12 UTC from IEEE Xplore. Restrictions apply.
4 TECHNICAL APPROACH fault position, respectively. Also the voltage boundary
conditions on bus k can be written as:
The two primary considerations is developing of an Vka, f = Z f 1 I ka, f + Z f 4 I k , f
effective solution technique for short circuit analysis are
Vkb, f = Z f 2 I kb, f + Z f 4 I k , f (9)
the using a suitable model to simulation fault on system
and selection an suitable method to calculate short Vk , f = Z f 3 I k , f + Z f 4 I k , f
c c

circuit parameters. In this paper, only a suitable model Where Vka, f , Vka, f and Vka, f are the voltages phases a, b
is used to simulation faults on network that is very
simple and useful. Also, the selected method to and c on fault condition respectively.
calculation short circuit parameters is based on
relationship matrices BIBC and BCBV according to After the fault occurs at bas k, the fault-bus currents will
their properties. flow and make the bus voltages of phases. Therefore,
the voltages variation after the fault can be expressed as:
4.1 Four Impedance Model ∆Vka, f = Vka, 0 − Z f 1 I ka, f − Z f 4 I k , f
Most of faults that occur on distribution system are ∆Vkb, f = Vkb, 0 − Z f 2 I kb, f − Z f 4 I k , f (10)
unsymmetrical faults. Unsymmetrical faults include ∆Vkc, f = Vkc, 0 − Z f 3 I kc, f − Z f 4 I k , f
single line (SLG), double line-to-ground (DLG), line-to-
line (L-L) and three line-to-ground etc. Usually, short Where Vkabc abc
, 0 and Vk , f are the pre-fault voltage and after-
circuit analysis methods for calculation fault parameter fault voltage, respectively. Substituting (7) into (4), the
use separated models for any type of faults that increase variation of the branch currents generated by the fault
the volume of calculations and memory requirements. In current can be expressed as:
this paper any types of unsymmetrical faults is
simulated by only one model that called four impedance
[ ] [
B f = [BIBC ] 0 ... I ka, f I kb, f I kc, f 0 T (11) ]
model [8]. As figure 2 shown, this model consists of Equation (11) can be written as:
four variable impedances that three impedances are on ⎡ I ka, f ⎤
three phases and another is located between phases and
ground.
[ ] [ ⎢
B f = BIBCka BIBCkb BIBCkc ⎢ I kb, f ⎥
⎥
] (12)
⎢ I kc, f ⎥
⎣ ⎦
Where BIBCkabc are the columns vectors of BIBC
matrix corresponding to three phases of bus k. By
substituting (12) into (6), the variation of the bus
voltages by the fault branch expressed as:
T
⎡ BIBCka ⎤ ⎡ I ka, f ⎤
⎢ ⎥ ⎢ ⎥
[ ∆V ] = [ BCBV ] ⎢ BIBCkb ⎥ ⎢ I kb, f ⎥ (13)
⎢ BIBCkc ⎥ ⎢ I kc, f ⎥
⎣ ⎦ ⎣ ⎦
Equation (13) can be formulated as:
T
Figure 2 Four impedance model. ⎡∆Vka, f ⎤ ⎡ BCBVka ⎤ ⎡ BIBCka ⎤ ⎡ I ka, f ⎤
⎢ b ⎥ ⎢ b⎥⎢ b⎥ ⎢ b ⎥
These impedances can be set from zero to extreme (for ⎢∆Vk , f ⎥ = ⎢ BCBVk ⎥ ⎢ BIBCk ⎥ ⎢I k , f ⎥ (14)
example 1e12). Therefore, any type of unsymmetrical ⎢∆Vkc, f ⎥ ⎢ BCBVkc ⎥ ⎢ BIBCkc ⎥ ⎢ I kc, f ⎥
faults can be modeled by setting these impedances ⎣ ⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦
suitably. For example, if Z f 1 and Z f 4 set be zero and Where BCBVkabc are the rows vectors of BCBV matrix
another impedances be 1e12, this model converted to corresponding to three phases of bus k. Equation (14)
single line-to-ground fault. Therefore each kind of faults can be written by substituting (8) and (10) into (14) as:
will be simulated by this model easily. ⎡Vka, 0 − Z f 1 I ka, f − ( I ka, f + I kb, f + I kc, f ) Z f 4 ⎤
⎢ b ⎥
⎢Vk , 0 − Z f 2 I k , f − ( I k , f + I k , f + I k , f ) Z f 4 ⎥ =
b a b c

4.2 Solution Techniques ⎢Vkc, 0 − Z f 3 I kc, f − ( I ka, f + I kb, f + I kc, f ) Z f 4 ⎥

According to proposed four impedance model, when a ⎣ ⎦ (15)
fault occurs on bus k as shown in fig.2, the current ⎡ L11 L12 L13 ⎤ ⎡ I ka, f ⎤
boundary conditions can be written as: ⎢ ⎥
⎢L
⎢ 21 L22 L23 ⎥⎥ ⎢ I kb, f ⎥
I ka = I ka, f , I kb = I kb, f , I kc = I kc, f (7) ⎢⎣ L31 L32 L33 ⎥⎦ ⎢⎣ I kc, f ⎥⎦
and for three phases connected to ground expressed as: T
⎡ L11 L12 L13 ⎤ ⎡ BCBVk ⎤ ⎡ BIBCk ⎤
a a
I k , f = I ka, f + I kb, f + I kc, f (8)
⎢ ⎥ ⎢ b⎥⎢ ⎥
Where [L] = L21 L22 L23 = ⎢ BCBVk ⎥ ⎢ BIBCkb ⎥ .
Where I ka, f , I kb, f and I kc, f are the currents phases a, b ⎢ ⎥
⎢⎣ L31 L32 L33 ⎥⎦ ⎢ BCBVkc ⎥ ⎢ BIBCkc ⎥
⎣ ⎦⎣ ⎦
and c and I k , f is total currents phases to ground at the
Equation (15) can be written as:

UPEC 2007 - 944

Authorized licensed use limited to: University of Moratuwa. Downloaded on September 07,2021 at 05:46:12 UTC from IEEE Xplore. Restrictions apply.
 faults on three buses are considered and the results are
shown in table 1.
(16)

p,:~o L/;j Table 1 Fault currents in three buses of IEEE 34-bus

test network.

Where [Z s ] - Zs4
Zf4
Therefore, the fault currents cab be calculated by:
Zs2 + Zs4
Zf4 ,41
Zs4
Zf3 -Jr-Zf4
B U s ~o
4
13
I SiigIe I~e I
A
900 A
272 A
a
1109 A
331 A
ee 0hise tO
b
1178 A
357 A
c
1151 A
349 A

k-';,,
,01
,
LV;o
20

Table 1 shows that the three phase fault currents are

more than single-phase to ground currents in each bus.
159 A 194 A 209 A 204 A

Where [Zsc] is the matrix related to the fault-bus voltage Also, fault currents for buses that are near to substation
and current at the fault position. Various types of faults are higher. For example, the fault current on bus 4 is
higher than buses 13 and 20.
can be simulated by changing Zsl to Zs4 that represent
in Z s matrix and then Zsc matrix respectively. After the Finding fault path currents is another useful property of
fault current at the fault position is calculated, the NEDA program. Figure 4 shows that the fault path
branch currents and bus voltages caused by the fault can current when a single-phase to ground fault occurs on
be calculated directly using 12 and 14, respectively. phase a of bus 4.
•u•u•u••u•••••••u••u••u•ùu•u•u••••u••••••u••••••u••u••u•••u••••u••u•••u•u•
i i iii iiii i
5 TEST CASE AND RESULTS

A software package called NEDA has been developed

by Delphi software to calculate the short circuit current
of unbalanced distribution networks. Different faults
conditions have been studied by using NEDA and the
results have been compared with the Digsilent
commercial software for validity of NEDA software.

Test network used in case studies is the modified IEEE ii~iiiiiiLi!~iiii~i~i!!!!iii~ii~iiii!iiiiiiiiii~iiii~iiiiiiiiii~!iiiii7~iii!!!iiiiiiiiiiii!iiiii]!~!!iii7~ii~i7ii~ii~!ii~iiii~Lii!~Cii

i ehiii!il eti~iifi~i7iii!i~ii!ii~!ii~i7i~i~!~i~ii!i!iii~i~i7i~ii!i!7ii~i~iii@iJi~ii~i!i!iii~i~ii!~i!~i:~i~i~i!i~ii?i~i7~i~i~ii~i~i~i~i~i~i7i!iiiiii~iii~iii~i~i!ii~i~ii~

34 bus radial distribution network [9], Fig. 3.

29 Figure 4 Fault path current of SLG fault on phase a.
28
In this figure, only branches that have been fault current
t 12
II 27 are shown. The loads along the fault path in NEDA
l0 24 I~ 26 program are omitted because they are assumed to have a
2 ] 4 6 7 £ 1] 14 3£ 31 32
_ -- - -- -- -- _ -- 2]
very small effect on the short circuit current. Therefore,
1 5 20
25

21
~
99
33
figure 4 shows the constant current on phase a only.

34 Case 2: The effect of fault on voltage profiles. The

voltage profiles for IEEE 34-bus test before fault is
15 16 17 18
shown in figure 5.
Figure 3 IEEE 34-bus test network.

Simplifying, the autotransformer 24.9/4.16 kV/kV in the

original IEEE 34-bus test feeder is replaced with the
line and the network is modelled with the single voltage
level. The automatic voltage regulator is also not
represented. The following cases are used to
demonstrate the proposed short-circuit analysis
program:

Case 1: The fault currents and the fault path in network.

Single-phase to ground fault is the most popular and
three-phase fault generally results in maximum short-
circuit current. Therefore, in this case, these types of
Figure 5 Voltage profiles before fault.

UPEC 2007- 945

Authorized licensed use limited to: University of Moratuwa. Downloaded on September 07,2021 at 05:46:12 UTC from IEEE Xplore. Restrictions apply.
Figure 6 shows that this feeder is unbalanced; therefore, been shown, the fault current magnitudes and the effect
an exact three-phase short circuit method can provide of them on voltage profiles depend on type and position
more accurate solutions. In this case, the effects of two of fault to substation. For example, three phases to
faults on the voltage profiles are analyzed. These faults ground fault have the highest current magnitude, so it
are: single line to ground on phase a of bus 4 and a line effects on whole of feeder more than another types.
to line between phase b that the effects of them are Also if the fault occurs near to substation, the fault
shown in figures 6 and 7 respectively. current and its effect on voltage profiles will be more
severe. Since the proposed method employs the actual
three-phase models for short circuit analysis and does
not require building the traditional admittance matrix,
the proposed method can achieve the advantages of high
speed, robust convergence and accuracy with lower
memory requirements especially.

7 REFERENCES

1. L. Roy, “Generalized polyphase fault-analysis

program:calculation of cross-country fault,” Proc.
IEE Vol. 126, No. 10, Oct. 1979.
2. W. H. Kersting, “Distribution system modeling and
analysis,” CRC Press, New York, 2002.
3. V. Brandwajn, W. F. Tinney, “Generalized method of
Figure 6 Voltage profiles of SLG fault on phase a. fault analysis,” IEEE Transaction on PAS, Vol. 104,
No. 6, pp. 1301-1306., June 1985.
4. X. Zhang, F. Soudi, D. Shirmohamdi. and C. Cheng,
“A distribution short circuit analysis approach using
hybrid compensation method,” IEEE Trans. Power
System., Vol. 10, No. 4, pp. 2053-2059, Nov 1995.
5. T. H. Chen, M. S. Chen, W. J. Lee, P. Kotas, V.
Olinda, “Distribution system short circuit analysis-
rigid approach,” IEEE Trans. Power System., Vol.
7, No. 1, pp. 444 450, Feb. 1992.
6. J. Teng, “systematic short circuit analysis method for
unbalanced distribution systems,” IEE Proc. Gener.
Transm. Distrib., Vol. 152, no. 4, pp. 549-555. July
Figure 7 Voltage profiles of L-L fault on phase 2005.
b and c. 7. J.H. Teng, “A direct approach for distribution
system load flow solutions,” IEEE Trans., pp. 882–
Figure 6 shows that the single line to ground faults in an 887, 2003, PWRD-18, (3).
increase in voltage magnitude on the other two phases, 8. R. M. Ciric, L.F. Ochoa, A. Padilla-Feltrin and H.
especially at the same bus. The amount of voltage rise Nouri, “Fault analysis in four-wire distribution
depends upon the system’s X/R ratio and mutual networks,” Proc. IEE Gener. Transm. Distrib., Vol.
impedances between phases. In figure 7, the voltages 152, No. 6, Nov 2005, pp. 977-985.
magnitudes of fault phases are decreased but exactly the 9. A. Kazemi, “Electrical power system,” IUST, Tehran,
second part, 2003.
same after fault due to Z f = 0 .

AUTHOR'S ADDRESS
6 CONCLUSIONS
The first author can be contacted at
This paper introduced a method for performing short
circuit analysis in unbalanced distribution networks. Electrical Engineering Department
Two matrices and a simple model are developed to Iran University of Science & Technology
analyze the variations of bus voltages, bus current Narmak
injections and branch currents under fault conditions. Tehran
From the above test results, it can be seen that the Iran
proposed algorithm can be used to handle Email [email protected]
unsymmetrical faults efficiently. As the results have

UPEC 2007 - 946

Authorized licensed use limited to: University of Moratuwa. Downloaded on September 07,2021 at 05:46:12 UTC from IEEE Xplore. Restrictions apply.