Branch Outage Simulation for

Contingency Studies
Dr.Aydogan OZDEMIR, Visiting Associate Professor
Department of Electrical Engineering,
Texas A&M University, College Station TX 77843
Tel : (979) 862 88 97 , Fax : (979) 845 62 59
E-mail : ozdemir@ee.tamu.edu

Aydoan zdemir was born in Artvin, Turkey, on

January 1957. He received the B.Sc., M.Sc. and
Ph.D. degrees in Electrical Engineering from
Istanbul Technical University, Istanbul, Turkey in
1980, 1982 and 1990, respectively. He is an
associate professor at the same University. His
current research interests are in the area of electric
power system with emphasis on reliability analysis,
modern tools (neural networks, fuzzy logic, genetic
algorithms etc.) for power system modeling,
analysis and control and high-voltage engineering.
He is a member of National Chamber of Turkish
Electrical Engineering and IEEE.

Power System Security

Power system security is the ability of the system to withstand one or more component
outages with the minimal disruption of service or its quality.

Outages of component(s)

Overstress on the other components

No limit violation

limit violation(s)

operation of protective devices

and switching of the unit(s)

partial or total loss of load

POWER SYSTEM
SECURITY

monitoring
contingency analysis
security constrained opf

Monitoring : Data collection and state estimation

The objective of steady state contingency analysis is to
investigate the effects of generation and transmission
unit outages on MW line flows and bus voltage
magnitudes.

START

SET SYSTEM MODEL TO

INITIAL CONDITIONS

SIMULATE AN OUTAGE OF A
GENERATOR OR A BRANCH

SELECT A
NEW OUTAGE

LIMIT VIOLATION
Y
ALARM MESSAGE
N
LAST OUTAGE
Y

END

Real-time applications require fast and reliable computation methods due to the high number of
possible outages in a moderate power system.
However, there is a well-known conflict between the accuracy of the method applied and the
calculation speed.

Exact solution

Full AC power flow

for each outage

not feasible
for real-time
applications.

Check the limit

violations

approximate methods to quickly

identify conceivable contingencies
real-time applications

AC power flows only for

critical contingencies.

Check the limit violations

APPROXIMATE CONTINGENCY ANALYSIS

Contingency ranking

contingencies are ranked in an approximate order of a scalar performance

index, PI.
contingencies are tested beginning with the most severe one and
proceeding down to the less severe ones up to a threshold value.
Masking effect causes false orderings and misclassifications.

Contingency screening

Explicit contingency screening is performed for all contingencies, following

an approximate solution (DC load flow, one iteration load flow, linear
distribution or sensitivity factors etc.)
Contingency screening is performed in the near vicinity of the outages (local
solutions)

Hybrid methods utilizing both the ranking and the screening

outage of a branch or a generation unit

MW line flow overloads

DC load flows
Sensitivity factors

voltage magnitude
violations

both

involves more complicated models

and better computation algorithms

LINE OUTAGE SIMULATION

An outage of a line can either be simulated by setting its impedance, yij = 0 or by injecting
hypothetical powers at both ends of the line. The latter method is preferred to preserve the
original base case bus admittance matrix.
i

S ij 0

yi0 0

y ij 0

Sij=0

S ji 0

y j0 0

Z-Matrix techniques
Modification of ZBUS is
required for each outage

Sji=0

S ij 0

S si

S ji 0j

y ij

yi0

y j0

S sj

Determination of the hypothetical sources so

that all the reactive power circulates through
the outaged line while maintaining the same
voltage magnitude changes in the system

SIMULATION FOR MW LINE FLOW PROBLEM

DC LOAD FLOW :
P B

[ B' ] ij 1 / x ij , [ B' ] ii 1 / x ik
k

, x ij Re al {1 / y ij }

outage of a line connected between busses i and j

P [ 0 0 .. 0 P si 0... P si 0..0]
X [ 0 0 .. 0 1 0 0 .. 1 .. 0 0 ]

; P si Re al { S si }

Psi , X [ B ]

The new real power flow through the line connected between busses n and m can be
derived and approximated as,
~
Pnm Pnm Pnm Pnm

1
x lm

([X] nn [X] mm - 2[X] nm ) Psi

See Power Generation, Operation and Control by Wood and Wollenberg for details

SIMULATION FOR VOLTAGE MAGNITUDE PROBLEM

Linear models are not sufficient for most outages
Reactive power flows can not be isolated from bus voltage phase angles

Involves more complicated models and better computation algorithms

i

Qij

Q ij

Q Li

Qji

ji

Lj

*
2
2 bi 0
Q ij Im ag { V i . y ij .V j } [V i V i V j cos ji ] b ij V i V j g ij sin ji V i

Transferring reactive power

T

assumed to flow through

the line

Q ij [V i V j ] b ij / 2 V i V j g ij sin ji
T

Q ji Q ij

Loss reactive power

Q
Q

[V i V j 2 V i V j cos ji ]
Li
Lj

Li

b ij
2

(V i V j )

bi0
4

assumed to allocated
at the busses

Can be split up
into two parts,

Line outage simulation by hypothetical reactive power sources

Q ij 0

Q ij

Q si Q ij Q Li

Q Li

Q ij

Q Li

Q ji 0

Q si Q ij Q Li

For a tap changing transformer, cross flow through the equivalent impedance is considered to be the
transferring reactive power, where shunt flows can be considered as the loss reactive powers.
bus i

a :1

bus i

bij

Q ij

bus j
1 1
( 1 ) b ij
a a

Q Li

bij

bus j

Q ji

Q Lj

(1

1
a

) b ij

Transferring reactive power is sensitive both to bus voltage magnitudes and bus voltage phase angles.
However, loss reactive power is dominantly determined by bus voltage phase angles and has a weak
coupling with bus voltage magnitudes. Therefore, transferring reactive powers are enough for a
reasonable accuracy.

Hypothetical reactive power injections to bus i and bus j, will result in a change in net
reactive bus powers Qi and Qj. This in turn, will result in a change in system state
variables with respect to pre-outage values. This change must be equivalent to the
changes when the line is outaged.
Load bus reactive powers do not satisfy the nodal power balance equation due to the
errors in load bus voltage magnitudes calculated from linear models. Therefore, part
of the fictitious reactive generation flows through the neighboring paths instead
circulating through the outaged branch. These reactive power mismatches can
mathematically be expressed as,
*

Q i Im ag V i Y ik V Q
Q Q si Q Di
k
ik
ij

k j
k
*

Q j Im ag V j Y jk V Q
Q
Q sj Q Dj
k
jk
ji

ki
k

where Qi and QDi are the net reactive power and the reactive demand at load bus i, is the
complex voltage at bus i and Yik is the element of bus admittance matrix. The superscript *
denotes the conjugate of a complex quantity. Calculated load bus voltage magnitudes need to
be modified in a way to minimize the bus reactive power mismatches at both ends of the
outaged line.
This can be accomplished a local optimization formulation

1. Select an outage of a branch, numbered k and connected between busses i and j.

2. Calculate bus voltage phase angles by using linearized MW flows.
l l ( X li X lj ) P k
Pk

, l=2,3,, NB

P ij
1 ( X ii X jj 2 X ij ) / x k

where X is the inverse of the bus suseptance matrix, Pij is the pre-outage active
power flow through the line and xk is the reactance of the line.

3. Calculate intermediate loss reactive powers,

~
Q

Li

~
Q

Lj

4. Minimize reactive power mismatches at busses i and j, while satisfying linear reactive
power flow equations. Mathematically, this corresponds to a constrained optimization
process as,
Minimize
wrt

T
Q ij

Subject

( Q i Q ij Q Di ) ( Q j Q ji Q Dj )

to g q ( V ) Q B V

~
T
Q ij Q ij Q Li
~
T
Q ji Q ij Q Li

reactive power flows

through the outaged
line

SOLUTION OF THE CONSTRAINED OPTIMIZATION PROBLEM

After having formulated the outage simulation as a constrained optimization problem,
minimization can be achieved by solution of the partial differential equations of the
augmented Lagrangian function
T

L ( Q ij , V , ( Q i Q ij Q Di )

( Q j Q ji Q Dj )

[B

Q V]

with respect to Q ijT , V and . Note that V does not need to include all the load bus
voltage magnitudes; instead only busses i, j and their first order neighbors are enough
for optimization cycle.

Drawback :

Convergence to local maximum

Single direction search

SOLUTION BY GENETIC ALGORITHMS

Evolutionary algorithms are stochastic search methods that mimic the metaphor of natural biological
evolution.
Genetic Algorithms (GAs) are perhaps the most widely known types of evolutionary computation methods
today.
GAs operate on a population of potential solutions applying the principle of survival of the fittest procedure
better and better approximation to a solution. At each generation, a new set of better approximations is created
by selecting individuals according to their fitness in the problem domain. This process leads to the evolution
of populations of individuals that are better suited to their environment than the individuals that they were
created from.
Generate initial
population
evaluate objective
function
GENERATE NEW
POPULATION

selection
crossover
mutation

optimization
criteria
met
Y
best
individuals
result

For the details of the processes see

Cheng, Genetic
Algorithms&Engineering
Optimization by M. Gen, R., New
York: Wiley, 2000 . Such a single
population GA is powerful and
performs well on a broad class of
optimization problems.

BASE CASE LOAD FLOW

SELECT AN OUTAGE
bounded network
CALCULATE BUS VOLTAGE PHASE ANGLES

j
Minimize

outaged branch

wrt
subject

Q ij

Q ji

Q ij

to V X Q

CALCULATE THE
REMAINING QUANTITIES

END

NUMERICAL EXAMPLES
IEEE 14-Bus test System
G

G
3

G
1
5

G
8

G
6

11

10

Base case control variables :

PG2 = 0.4 p.u.
PG3 = PG6 = PG8 = 0.0 p.u.
V1 = 1.06 p.u.
V2 = 1.045 p.u.
V3 = 1.01 p.u.
V6 = 1.07 p.u.
V8 = 1.09 p.u.
B9 = 0.19 p.u.
t4-7 = 0.978
t4-9 = 0.969
t5-6 = 0.932

12

13

14

Q7-9 = 27.24 Mvar

Q5-6 = 12.42 MVar

P o s t-O u ta g e V o lta g e M a g n itu d e s fo r I E E E -1 4 B u s T e s t S y s te m

B us

O u ta g e o f L in e 7 -9

O u ta g e o f tra n s fo rm e r 5 -6

N o V L F [p u ] V P F [p u ] V [% ] V L F [p u ] V P F [p u ]

V [% ]

1 .0 6 0

1 .0 6 0

0 .0

1 .0 6 0

1 .0 6 0

0 .0

1 .0 4 5

1 .0 4 5

0 .0

1 .0 4 5

1 .0 4 5

0 .0

1 .0 1 0

1 .0 1 0

0 .0

1 .0 1 0

1 .0 1 0

0 .0

1 .0 1 5

1 .0 1 5

0 .0

1 .0 1 5

1 .0 2 3

0 .8

1 .0 1 6

1 .0 1 8

0 .2

1 .0 2 5

1 .0 3 2

0 .7

1 .0 7 0

1 .0 7 0

0 .0

1 .0 7 0

1 .0 7 0

0 .0

1 .0 6 6

1 .0 6 8

0 .1

1 .0 5 5

1 .0 5 5

0 .0

1 .0 9 0

1 .0 9 0

0 .0

1 .0 9 0

1 .0 9 0

0 .0

0 .9 8 8

0 .9 9 3

0 .5

1 .0 4 6

1 .0 3 8

0 .8

10

0 .9 9 4

0 .9 9 9

0 .5

1 .0 4 3

1 .0 3 6

0 .7

11

1 .0 2 7

1 .0 3 0

0 .3

1 .0 5 3

1 .0 4 9

0 .4

12

1 .0 5 0

1 .0 5 1

0 .1

1 .0 5 2

1 .0 5 4

0 .2

13

1 .0 4 0

1 .0 4 1

0 .1

1 .0 4 9

1 .0 4 8

0 .1

14

0 .9 9 2

0 .9 9 6

0 .4

1 .0 2 8

1 .0 2 4

0 .4

M a x im u m e rro r: 0 .5 %

M a x im u m e rro r: 0 .8 %

Post-outage reactive power flows for IEEE-14 Bus Test Systems

Line
l=m

1-2

Outage of Line 7-9

Outage of transformer 5-6
QPF
QDF
QPF
QDF
Q
Q
[MVa [Mvar
[MVar]
[Mvar]
[Mvar]
[Mvar]
r]
]
-20.3 -20.2
0.07 -21.6 -21.1
0.53

1-5

5.4

4.4

2-3

3.6

3.6

2-4

0.2

-0.1

2-5

2.8

1.7

3-4

5.3

5.0

4-5

12.0

9.0

4-7

-14.1

-14.8

4-9

13.2

12.9

5-6

12.8

13.8

6-11

14.6

12.9

6-12

3.7

3.5

6-13

13.0

12.0

7-9

0.98
0.02
0.27
1.15
0.33
3.02
0.70
0.32
0.97
1.73
0.20
0.96

86.7

9-10

-5.5

-4.8

9-14

-2.6

-1.9

10-11

-11.3

-10.2

12-13

1.9

1.6

13-14

8.3

7.4

7-8

-14.5

-13.3

0.71
0.70
1.11
0.34
0.85
1.21

1.3
3.3
-1.6
-1.3
3.7
8.6
-5.1
3.0
19.5
5.1
15.1
9.6
-8.2
-4.6
-14.9
3.4
12.4
-21.2

-1.3
3.3
-5.8
-4.2
-0.1
14.0
-0.8
6.4
42.6
19.9
4.7
15.5
17.7
-8.9
-5.5
-15.5
3.5
12.2
-21.2

2.64
0.03
4.15
2.90
3.81
5.35
4.31
3.35
0.41
0.36
0.42
8.12
0.66
0.88
0.64
0.06
0.24
0.04

IEEE 57-Bus Test System

5

16

45
18

19

15

17

14

13

20
21

12

46
47
44 48

26

50

24

49
23

38

22

39

37

25

40

56

57
41

11

36
27

28

29

35

30

33

31

32

34

53

54

52

42

43

55

10

51

First one is the outage of the line connected between bus-12 and bus-13, whose preoutage reactive power flow is 60.27 Mvar. Second case is the outage of a transformer
with turns ratio 0.895 connected between bus-13 and bus-49, whose pre-outage reactive
power flows is 33.7 Mvar.
Post-Outage Voltage Magnitudes for outage of the line connected between bus 12 and bus
V
Bus No Voltage magnitudes [p.u.]
pre-outage VPF
VDF
13
0.979 0.955 0.953 0.0019
14
0.970 0.953 0.951 0.0018
20
0.964 0.955 0.953 0.0016
46
1.060 1.042 1.040 0.0023
47
1.033 1.016 1.014 0.0016
48
1.028 1.011 1.009 0.0020
49
1.036 1.019 1.017 0.0024
threshold error = 0.0015 p.u.

Post-Outage Reactive Power Flows for outage of the

line connected between bus 12 and bus 13

Reactive Power Flow [MVar]

Line pre-outage
l-m Qlm Qml
1-2
1-15
3-15
50-51

75.00 -84.12
33.74 -23.95
-18.26 13.73
-4.16 6.51

QPF
Qlm

QDF
Qml

Qlm Qml

74.84 -83.94 75.01 84.14

45.29 -34.96 46.26 35.22
0.54 -5.15 0.87 -5.26
-9.43 9.92 -9.23 9.78
threshold error = 0.2 MVar.

Q
[MVar]

0.17
0.97
0.33
0.20

0.20
0.26
0.11
0.14

Post-Outage Voltage Magnitudes for outage of the transformer

connected between bus 13 and bus 49

Bus No
11
13
21
48
49
50
51

Voltage magnitudes [p.u.]

pre-outage
VPF
VDF
0.974
0.976
0.977
0.979
0.985
0.987
1.009
0.982
0.980
1.028
0.997
0.995
1.036
0.978
0.972
1.024
0.980
0.977
1.052
1.038
1.036
threshold error = 0.0015 p.u.

V
0.0011
0.0016
0.0017
0.0016
0.0056
0.0032
0.0018

Post-Outage Reactive Power Flows for outage of the transformer

connected between bus 12 and bus 13

Reactive Power Flow [MVar]

Line
l-m
3-15
12-13
15-45
14-46
47-48
48-49
50-51
10-51

pre-outage
Qlm
Qml
-18.26 13.73
60.27 -64.01
-0.79 2.15
27.32 -25.39
12.36 -12.26
-7.40 6.95
-6.16 6.51
12.47 -11.81

QPF

QDF

Qlm
Qml
Qlm
Qml
-15.59 11.01 -17.09 12.53
52.49 -56.76 50.06 -54.46
7.67 -5.67 9.33 -7.36
42.82 -39.29 45.93 -42.24
24.76 -24.41 22.71 -22.27
5.93 -6.10 4.31 -4.20
-13.25 14.53 -11.84 13.35
21.06 -19.83 23.24 -21.98
threshold error = 1.0 MVar.

Q
[MVar]

1.50
2.43
1.66
3.11
2.05
1.62
1.41
2.18

1.52
2.30
1.69
2.95
2.14
1.90
1.18
2.15