ATC Based Optimal Power Flow for Contingent

Power Systems Using Differential Evolution

Chanakan Cholsuk#, Padej Pao-La-Or#, Thanatchai Kulworawanichpong#
#
School of Electrical Engineering, Institute of Engineering, Suranaree University of Technology,
NaKhon Ratchasima, Thailand 30000

1
chanakan_na@hotmail.com
2
padej@sut.ac.th
3
thanatchai@gmail.com

As mentioned, it is necessary to consider the

Abstract In this paper, optimal power flow based on available
transfer capability (ATC) has been proposed for contingent
optimal operating point after some equipment
power systems. In general, the total production cost is commonly
outage occurs in the power system. The planning
used as the main objective for optimal power flow problems.
must be done before an actual outage circumstance
However, in some circumstances, where the economical point of
views is less important such as under an emergency state of
is really happened. Contingency analysis of power
partial blackout, the transmission security objective e.g. ATC is
systems [1] is a powerful tool to be done under this
very useful. Maximizing the ATC implies to reduce probability of
assumption. To increase reliability and safety of the
wide area blackout as a consequence. This paper performs
contingency analysis of the power system. After an occurrence of
power system, contingency analysis in conjunction
some specific component failure, an ATC based optimal power
with ATC based optimal power flow has been
flow problem has been formulated and solved by differential
proposed in this paper in order to maximize the
evolution (DE) method. To illustrate its use, the standard IEEE
14-bus test system was tested.
transmission system security.
In this paper, six sections are organized. Section
Keywords Available Transfer Capability , Contingency
analysis, Performance index, fuel cost, power losses
II illustrates the description of contingency analysis
with optimal power flow based ATC. Section III
I. INTRODUCTION reviews the theoretical background of differential
The current power systems are large, complex evolution as a powerful searching method.
and power transfer quantities. Therefore, the power Simulation results are discussed in Section IV.
systems should be considered safe for man, Section V, the last section, provides the conclusion.
equipment and environment. There were several
severe blackout occurred somewhere in the world
II. CONTINGENCY ANALYSIS WITH ATC-BASED OPF
including Thailand. The case of a well-known
power outage throughout New York City in August The contingency occurs for any of collapsing
14, 2003, affected about 40 million people. Later, one by one or multiple devices. In this paper, the
the power outage in Italy was dated on September first-order contingency is only considered. After the
28, 2003 and this was caused by a power failure in situation occurs it is necessary to find an optimal
France and Switzerland. The 2003 Italy blackout operation under this abnormal condition to prevent
resulted in power outage for about 15% of the a consecutive equipment outage that might cause
whole country and approximately 57 million people wide-area or complete blackout of the entire power
experienced this effect. In Thailand, the serious system.
blackout was taken place over three decades ago Since the contingency analysis is based on
(March 18, 1978). This blackout was caused by the simulation of power system operation under some
failure of the generator at South Bangkok Power equipment failure, it is a time consuming process to
Plant. The whole area blackout can probably be complete this study. This process can be
happened in future if efficient power system summarized as shown in Fig 2.
planning has not been done.

978-1-4799-0545-4/13/$31.00 2013 IEEE

 line outage distribution factor, active power
START performance index, reactive power performance
index, etc, can be employed.
In this paper, the optimal power flow problem
under some equipment failure has been studied. The
objective function for this case is different from that
Load system data of the normal OPF problem. Available transfer
capability (ATC) is chosen as the objective.
According to the report of NERC (1995) [2],
transfer capability refers to the ability of
transmission systems to reliably transfer power
Count i=1 from one area to another over all transmission paths
between those areas under given system conditions.
The mathematical definition of ATC given in the
report of NERC (1996) is . . . the Total Transfer
Capability (TTC) less the Transmission Reliability
Apply equipment
Margin (TRM), less the sum of existing
failure transmission commitments and the Capacity Benefit
Margin (CBM): that is

ATC = TTC - TRM - ETC (1)

Find optimal operation by optimal
power flow based ATC Power Flow (MW)

TTC

NO
i=i+1 Terminate? ATC

YES
ETC
Record results

Fig. 3 Available Transfer Capability

The evaluation of ATC can be formulated as an

STOP
optimization problem. The objective function to be
maximized is expressed as (2) and the optimization
Fig. 2 Flow diagram representing contingency analysis
problem is subjected to some constraints.
There are thousands of possible outages for
contingency analysis to be studied. This can cause Maximize ATC (2)
lengthy time to complete the task. One of the
easiest ways is to provide the quick calculation that To solve this optimization problem, the efficient
can evaluate transmission security performance. method of differential evolution is exploited to seek
Some sensitivity factors e.g. generation shift factor, for optimal solutions.
 III. DIFFERENTIAL EVOLUTION

According to the description by Storn and Price

[3], the classical differential evolution algorithm
can be outlined in the following steps (see also in
Fig. 4).
A. Initialization
Create answer to the decision variables.( Xi )
There are NP sets. The decision variables, D. The
calculation of the objective function of each
answer.
B. Mutation
Create tangent vectors of NP sets. ( target vector,
Xi,G variables, D) 3 random vector that is unique to
the tangent vectors. (Xr1,G, Xr2,G, Xr3,G) Mutation
(Mutation. Vi,G+1) by using the equation.(3)

Vi,G+1 = Xr1,G + F(Xr2,G Xr3,G) (3)

F is weighing factor between 0 to 2

Fig. 4 Flow diagram of differential evolution
C. Crossover
The crossover Answer varied according to Equation (4). IV. SIMULATION RESULTS AND DISCUSSION
In this paper, the standard IEEE 14-bus test
Uji,G+1 = (U1i,G+1, U2i,G+1, U3i,G+1,, UDi,G+1) system [4] was used for test as shown in Fig 5.
V ji ,G +1 if (randb ( j ) CR or j = rnbr (i ) (4)
U ji ,G +1 =
X ji ,G if (randb ( j ) > CR or j rnbr (i )
where
Uji,G+1 = Trial vector
Xji,G+1 = Mutant vector
Vji,G = Target vector
randb(j) = is a randomly chosen index to ensure
that at least one of the variables should be
changed between 0-1.
CR = Crossover Constant There is a real
number between 0-1.
rnbr(i) = is the index of the random integer Fig. 5 IEEE 14-bus test system
value between 0 D-1.
In this paper, optimal operation of contingent
D. Selection power systems based on ATC objective
The objective function values obtained from trial consideration was investigated. The simulation was
vector (Ui,G+1 ) and target vector (Vi,G+1 ) . Vector conducted by performing the first-order
that gives a better answer than to be stored. Repeat contingency by the outage of equipment. The
steps 2 through 4. transmission line outage can be considered as 23
separate cases, while generator outages and Loss 832.71 3.9400 1107.5 4.0959
transformer outages are 4 and 3 cases respectively.
ATC 1153.6 4.6286 1116.1 6.3522
Due to the limit of spaces, only four cases of the
transmission outages were discussed (case 1 4, PI 1154.6 4.6018 1113.3 6.3771
3
see Fig. 5). When applying Case 1 for line 1-2 Cost 944.76 5.4410 1003.1 8.3754
outage the optimal solutions for optimal power flow
Loss 870.52 4.7836 1.097.0 5.3705
problem can be presented in Table I. The optimal
value of control parameters is dependent on which ATC 1202.2 4.2453 1107.4 4.1579
the objective function being used. The optimal PI 1198.0 4.1448 1099.3 4.1351
results of the four cases are summarized in Table II. 4
Cost 908.30 4.9793 992.95 6.2901

TABLE I Loss 833.62 4.2687 1107.1 3.7332

OPTIMAL SOLUTION FOR CASE 1 (LINE 1-2 OUTAGE)

case 1 (line 1-2 outage)

Objective function V. CONCLUSIONS
variable
ATC PI Cost Loss This paper proposed the ATC based optimal
V2 1.0996 1.0596 0.9924 1.0568
V3 1.0812 1.100 0.9787 1.0873 power flow for contingent power systems using
V6 1.0188 0.9246 1.0337 1.0880 differential evolution. When a single component
V8 1.0993 1.0254 1.0602 1.0818 outage occurs the system must be operated at a safe
del2 -0.1300 -0.1358 -0.3468 -0.1394
del3 -0.1925 -0.1950 -0.3982 -0.1924
and secure operating point. This paper employed
del6 -0.1758 -0.1782 -0.3486 -0.1860 the IEEE 14-bus test system to distinguish the
del8 -0.0929 -0.1068 -0.2765 -0.1160 differences among four objective functions. The
T1 1.0203 0.9000 1.0888 0.9496 results showed that the ATC based optimal power
T2 0.9030 0.9756 0.9111 0.9017
T3 1.0816 0.9001 0.9742 0.9082
flow can guarantee a safe and secure operating
PG2 79.9685 79.8737 22.5578 79.2534 condition under a specific component outage.
PG3 49.9897 49.8684 43.1760 48.8879
PG6 29.9083 29.8992 29.9961 29.9982
PG8 34.9767 34.9563 34.9995 34.3586 REFERENCES
QG2 49.4836 59.9267 19.3755 44.4658 [1] A.J. Wood and B.F. Wollenberg, Power generation, operation and
QG3 13.6888 12.1976 36.1335 27.3331 control, Wiley-Interscience, 1996.
QG6 13.1103 61.5353 11.2042 20.5677 [2] Available Transfer Capability Definitions and Determination, North
American Electric Reliability Council, Reference Document, June
QG8 30.0581 29.8177 8.8681 33.7794
1996. Source: www.westgov.org/wieb/wind/06-96NERC atc.pdf.
objective value 1127 4.0928 1023 4.3555 [3] R. Storn and K. Price, Differential evolution A simple and efficient
heruistic for global optimization over continuous spaces, Journal of
Global Optimization, vol.11, pp.341-359, 1997.
TABLE II [4] H. Sadaat, Power system analysis, McGraw-Hill, 2002.
SUMMARY OF THE FOUR TEST CASES

First-order contingency

Objective Performance
Case
function ATC PI Cost Lose

ATC 1127.4 4.1892 1109.6 4.3692

PI 1124.8 4.0670 1114.8 7.0531

1
Cost 1011.3 5.5243 993.6 6.4553

Loss 887.75 4.2002 1107.6 4.0457

ATC 1201.5 3.9368 1107.9 4.2948

2 PI 1200.5 3.9242 1106.1 4.3720

Cost 912.24 4.6292 999.64 7.9754