Académique Documents
Professionnel Documents
Culture Documents
Abstract
Contingency
analysis
(CA)
is
an
important
tool
in
the
Energy
Management
System
both
for
design
and
operation
of
a
power
network.
It
analyses
the
system
security
and
the
need
for
a
remedy
in
case
of
a
fault
in
the
network.
This
paper
performs
(N-1),
(N-2)
and
(N-3)
contingency
analysis
on
two
standard
IEEE
test
cases
and
shows
the
difficulty
of
scaling
this
analysis
for
a
higher
(N-x)
given
the
exponential
increase
in
time
take
between
different
levels.
Keywords
-
Contingency
Analysis,
Power
System
Stability,
Power
Flow
I.
Introduction
Contingency
analysis
(CA)
is
a
key
component
of
operating
todays
energy
management
system
(EMS).
It
is
used
to
tell
operators
what
might
happen
in
case
of
an
equipment
outage.
It
analyses
whether
the
system
can
still
operates
within
limits
upon
facing
an
outage
or
whether
the
under
voltages
and
overloads
require
a
remedy.
It
is
key
part
of
what
the
operator
does
to
avoid
equipment
damage
and
to
minimize
customer
outages.
i
II.
Motivation
The
security
of
a
power
grid
has
come
into
increased
focus
given
the
emerging
concerns
around
the
threat
to
a
power
network
from
terrorism.
A
recent
report
widely
circulated
in
the
national
media
concluded
that
the
U.S.
could
suffer
a
national
blackout,
if
only
nine
out
of
fifty
five
substations
were
knocked
out.
ii
This
underscores
the
importance
of
contingency
analysis,
which
calculates
power
system
stability
upon
disabling
some
parts
of
the
network.
III.
Typical
Causes
of
Faults
The
most
common
cases
of
contingencies
on
a
power
network
involve
equipment
failure,
either
a
loss
of
a
generating
unit,
or
a
loss
of
a
transmission
component
including
a
transmission
line,
transformer,
substation
bus
or
switching
gear.
Furthermore
failure
can
be
single
or
multiple
if
relay
failures
are
taken
into
account.
The
other
common
causes
of
faults
are
weather
related
and
the
distribution
of
causes
varies
by
utility
(Figure
1.0).
iii
simplicity
for
initial
analysis.
The
IEEE
300
was
chosen
to
see
how
the
analysis
would
translate
to
larger
networks,
with
more
complexity.
The
example
codes
for
the
simulations
can
be
found
in
the
appendix.
VI.
Results
for
IEEE
Case
30
This
case
represents
a
real
network
that
was
a
portion
of
the
American
Electric
Power
System
(in
the
Midwestern
US)
as
of
December
1961.
It
has
30
buses
and
41
branches.
A
drawback
of
this
case
is
that
it
doesnt
have
line
limits.
A
bus
diagram
representation
can
be
seen
in
Figure
2.0.
(N-1)
Contingency
analysis
of
this
network
was
performed
by
iteratively
switching
off
one
bus
for
each
of
the
41
buses
and
checking
for
convergence.
The
analysis
revealed
that
convergence
was
not
found
upon
switching
off
the
13th,
16th
or
34th
branches.
These
branches
correspond
to
the
connections
from
bus
9
to
11,
12
to
13
and
25
to
26
respectively
(see
Figure
2.0).
This
intuitively
makes
sense
as
the
first
two
cases
disconnect
the
synchronous
generator
and
the
last
case
is
connected
to
a
load
with
a
single
line
and
no
redundancy.
Figure
2.0
IEEE
Test
Case
30,
with
the
failure
nodes
for
N-1
CA
circled
(N-3)
Contingency
analysis
of
this
network
was
performed
by
iteratively
disabling
three
branches
of
the
network
at
a
time
and
looking
for
convergence,
the
results
of
which
are
seen
in
Figure
4.0.
The
figure
shows
are
41X41X41
matrix
with
each
cell
representing
the
success
or
failure
of
convergence.
Yellow
represents
success,
while
red
represents
failure
(Note
opposite
coloring
scheme
from
the
N-2
case).
The
matrix
is
again
completely
symmetric,
as
expected.
To
get
a
more
intuitive
understanding
of
what
is
happening
we
can
sum
over
the
number
of
failures
on
the
3rd
dimension.
The
results
of
this
are
seen
in
Figure
5.0.
We
can
see
that
we
have
an
(N-3)
failure
whenever
we
have
an
(N-2)
failure.
The
situations
in
which
we
would
not
have
had
a
failure
if
one
of
the
branches
were
active
can
be
seen
in
the
undulations
next
to
the
origin
where
we
have
about
6-7
failures
for
a
particular
entry
of
the
matrix.
Figure
5.0
(N-3)
CA
on
IEEE
Test
Case
30,
with
3rd
Dimension
Summed
Over
Table
1.0
Results
of
the
CA
for
IEEE
Case
30
and
time
taken
for
the
simulation
IEEE
CASE
30
Number of iterations
Time(s)
Number of Failures
Failures/Iterations Time/Iterations
N-1
41
1.678
7.32%
0.040926829
N-2
1681
65.233
289
17.19%
0.038806068
N-3
68921
3360.781
19797
28.72%
0.048762801
VII.
Results
for
IEEE
Case
300
This
case
was
made
by
Mike
Adibi
in
1993
for
the
IEEE
Test
Systems
Task
Force.
It
has
300
buses
and
411
branches.
Unlike
Case30
it
has
line
limits,
which
is
part
of
the
reason
we
shall
a
higher
failure
rate
for
this
system.
The
motivation
for
choosing
this
case
was
to
see
how
CA
scales
with
an
increase
in
number
of
buses.
(N-1)
Contingency
analysis
was
performed
by
iteratively
disabling
each
of
the
411
buses,
the
results
of
which
can
be
seen
in
Figure
6.0.
The
white
bands
represent
failure
and
the
blue
bands
represent
success.
The
significant
change
from
Case
30
is
that
the
failure
rate
is
much
higher
(25%)
and
the
time
taken
for
each
iteration
is
a
significant
5
times
larger
than
Case
30.
See
Table
2.0
for
time
taken
results.
Figure
6.0
Result
of
(N-1)
on
IEEE
Test
Case
300,
white
is
failure,
blue
is
success
(N-2)
Contingency
analysis
of
this
network
was
performed
by
iteratively
disabling
two
branches
of
the
network
at
a
time
and
looking
for
convergence,
the
results
of
which
are
seen
in
Figure
7.0.
The
figure
shows
are
411X411
matrix
with
each
cell
representing
the
success
or
failure
of
convergence.
Yellow
represents
success,
while
red
represents
failure.
As
expected,
the
matrix
is
symmetric
and
we
have
failure
in
(N-2)
whenever
we
have
failure
in
(N-1).
From
Table
2.0
we
see
that
number
of
iterations
increases
exponentially
and
(N-2)
requires
411^2
iterations.
Figure
7.0
Result
of
(N-2)
on
IEEE
Test
Case
300,
Red
is
failure,
Yellow
is
success
An
(N-3)
analysis
was
not
performed,
as
it
would
have
taken
a
very
large
amount
of
time.
(N-2)
analysis
required
41623
seconds
(~11.5
hours)
and
(N-3)
would
have
at
least
taken
17107253.57
s
(~4752
hours).
Note
also
the
significant
increase
in
Failure
%
and
increase
in
time
per
iteration
in
Table
2.0.
Table
2.0
-
Results
of
the
CA
for
IEEE
Case
300
and
time
taken
for
the
simulation
IEEE
CASE
300
N-1
Number of iterations
Time(s)
Number of Failures
Failures/Iterations Time/Iterations
411
81.664
105
25.55%
0.198695864
N-2
168921
41623.488
75603
44.76%
0.246408013
N-3
69426531
Conclusion
This
paper
performed
contingency
analysis
on
different
IEEE
test
cases.
The
major
limitation
in
continuing
this
analysis
was
the
amount
of
time
CA
takes
for
larger
systems
(with
more
buses)
and
to
perform
analysis
in
greater
depth,
i.e.
(N-x)
with
x>3.
This
is
due
to
the
exponential
time
it
takes
for
an
increase
in
the
depth
of
the
analysis.
Finding
a
better
algorithm
to
perform
this
is
an
active
area
of
research
with
efforts
such
as
particle
swarm
optimizationvi,
contingency
screeningvii
and
many
more.
There
is
even
an
area
of
study
on
how
unsolvable
there
problems
can
beviii.
While
this
paper
does
not
offer
a
solution
to
this
complex
problem,
it
shows
us
using
standard
cases
why
this
is
a
hard
to
problem
and
also
an
important
one.
Appendix
Example
code
used
to
perform
CA
(This
one
is
for
N-1
and
IEEE
Case
30)
(This
one
is
for
N-2
and
IEEE
Case
300)
References
i
R.
Bacher,
Graphical
Interaction
and
Visualization
for
the
Analysis
and
ii
Smith,
Rebecca.
"U.S.
Risks
National
Blackout
from
Small
Scale
Attack."
The
Wall
Street
Journal.
Dow
Jones
&
Company,
12
Mar.
2014.
Web.
15
May
2014.
iii
Brown,
Richard
E.
Electric
Power
Distribution
Reliability.
Boca
Raton,
FL:
CRC,
2009.
Print.
iv
G.
C.
Ejebe
and
B.F.
Wollenberg,
Automatic
Contingency
Selection,
IEEE
Trans.
on.
PAS-98,
pp.
97-109,
Jan/Feb
1979.
v
Power
Systems
Test
Case
Archive.
University
of
Washington,
Department
of
Electrical
Engineering,
n.d.
Web.
vi
H.
Yoshida,
K.
Kawata,
Y.
Fukuyama,
S.
Takayama,
and
Y.
Nakanishi
"A
particle
swarm
optimization
for
reactive
power
and
voltage
control
considering
voltage
security
assessment",
ibid.,
vol.
15,
pp.
1232-1239,
Nov.,
2000
vii
G.
C.
Ejebe,
G.
D.
Irisarri,
S.
Mokhtari,
O.
Obadina,
P.
Ristanovic,
and
J.
Tong,
"Methods
for
contingency
screening
and
ranking
for
voltage
stability
analysis
of
power
systems",
Proc.
PICA
Conf.,
pp.249
-255
1995
viii
T.
J.
Overbye,
"A
power
flow
solvability
measure
for
unsolvable
cases",
IEEE
Trans. Power Syst., vol. 9, pp.1359 -1365 1994