ADAPTIVE TUNED DIFFERENTIAL EVOLUTION BASED

VOLTAGE SECURITY CONSTRAINED OPTIMAL POWER FLOW

INCLUDING GENERATION RESCHEDULING
ABSTRACT
Voltage instability has become a major concern in planning and operation of many power
systems. Contingencies such as unexpected line outages in stressed system may often result in
voltage instability which may lead to voltage collapse. This paper deals with voltage stability
enhancement in power system through optimal power flow. The maximum L-index of the load
buses is taken as the indicator of voltage instability. An Adaptive Differential Evolution (ADE)
algorithm is proposed to solve the Voltage Security Constrained Optimal Power Flow (VSCOPF) problem. Real power generator settings, voltage magnitudes, transformer tap settings and
reactive power generation of capacitor banks are taken as control variables. DE/rand/1/bin
strategy scheme which employs binomial crossover and difference vector based mutation along
with adaptive tuned parameters is used to solve the VSCOPF problem. The effectiveness of the
proposed algorithm is demonstrated through IEEE 30-bus system and IEEE 57 bus test system
and the results are compared with the results of Genetic algorithm (GA). Simulation results show
the effectiveness of the proposed approach for improving the voltage stability of the system
under base case and contingency states.
Keywords: Optimal Power Flow, Voltage Stability, L-index, Generation Rescheduling,
Differential Evolution.
List of symbols
FT Total fuel cost;
P loss Network real power loss;
Pi, Qi Real and Reactive Powers injected into network at bus i;

Gij, Bij Mutual conductance and susceptance between bus i and bus j;
Gii,Bii Self- conductance and susceptance of bus i;
Pgi, Qgi Real and Reactive power generation at bus i ;
QCi

Reactive power generated by ith capacitor bank;

tk

Tap setting of transformer at branch k ;

Vi

Voltage magnitude at bus i;

Vj

Voltage magnitude at bus j;

ij

Voltage angle difference between bus i and bus j;

Sl

Apparent power flow through the lth branch;

gk

Conductance of branch k;

NB

Total number of buses;

NB-1

Total Number of buses excluding slack bus;

NPQ

Number of PQ buses;

Ng

Number of generator buses;

Nc

Number of capacitor banks;

NT

Number of tap-Setting transformer branches;

Nl

Number of branches in the system.

PGik

Real power generation in contingency state

PGi0

Real power generation in base case

Tk

720 secs

dPGi

dt

max

maximum rate of change of generation from base case to contingency state

I. INTRODUCTION
Optimal Power Flow (OPF) problem is one of the most widely used tools in planning and real
time operation in the modern Energy Management Systems. The OPF problem [1,2] aims to
achieve an optimal solution of a specific objective function such as fuel cost by adjusting the
power system control variables while satisfying a set of operational and physical constraints. A
number of mathematical programming based techniques have been proposed to solve the optimal
power flow problem. These include gradient method [1,2,3], Newton method [4], linear
programming [5,6] and Interior point method [7]. These traditional techniques rely on convexity
to find the optimum solution. But due to the non-linear and non-convex nature of the OPF
problem, the methods based on these assumptions do not guarantee to find the global optimum.
Also, the discrete variables related to the tap changing transformer and shunt capacitors cannot
be incorporated directly into the general optimal power flow programs. Recently Evolutionary
Computation Techniques like Genetic Algorithm [8] and Evolutionary Programming [9,10] have
been successfully applied to solve the OPF problems. Evolutionary computation techniques does
not require any space limitations such as smoothness, convexity or unimodality of the function to
be optimized. This feature makes it suitable for many real world applications including the OPF
problem.

In the recent past, several voltage instability incidents have been reported, all over the
world. This problem can only be exasperated by the application of open market principles to the
operation of power systems, as stability margins are being reduced even further to respond to
market pressures, which demand greater attention to reduced operating costs. This has
necessitated the importance of incorporating the voltage stability limit in the power system
operation and planning stage. For voltage security enhancement Voltage Security Constrained
Optimal Power Flow (VSC-OPF) [11-15] has been proposed. In VSC-OPF, voltage stability is
incorporated into the OPF problem to have enough voltage stability margin. Voltage instability
can be avoided by preventive and corrective controls [16]. The post-contingency corrective
control is carried out to improve the voltage stability performance after a severe contingency. In
preventive control action is taken before the occurrence of the contingency itself. It is a non
linear, non convex, large scale optimization problem with both continuous and discrete variables.
The integer variables appear in the mathematical formulation due to the discrete nature of the
transformer tap positions and the capacitor bank. The voltage stability problems cannot be fully
solved by reactive power resources and voltage control equipments in the system. Generation
rescheduling is also necessary to change the real power settings of the generators in contingency
state to improve the voltage stability of the system.
In [17], generation rescheduling was considered to change the system operating
conditions to ensure voltage stability under contingency state using multi contingency sensitivity
based approach. Wu et al [18] proposed an effective scheme for rescheduling pool generation
and for adjusting contract transactions to ensure adequate voltage stability margin. Canizares et
al [14] incorporated voltage stability into OPF problem minimizing generation cost, transmission

losses and improving voltage security of the system. Devaraj et al [19] proposed an improved
Genetic Algorithm to solve the VSC-OPF problem.
In this paper, contingency state voltage stability level is incorporated through L-index
[20] into the OPF problem, to ensure voltage security of the system. In contingency state, by
including Lmax in the objective function, voltage stability is achieved at the expense of operating
costs. This increase in operating costs can be reduced by generation rescheduling in contingency
states. The generator ramp rates can restrict the speed with which active power is rerouted in the
network. Hence, during rescheduling of generators in contingency state, generator ramp rate is
incorporated as an additional constraint of the VSC-OPF problem.
In this work, an Improved DE [21] with self tuned parameters is applied to solve the
ramp rate constrained VSC-OPF problem. In the proposed DE, mutation scheme uses a randomly
selected vector and only one weighted difference vector is used to perturb it. The mutation
scheme is combined with binomial type crossover and with random scalar vector. The
effectiveness of this algorithm is demonstrated through voltage stability constrained optimal
power flow problem in IEEE 30-bus system and IEEE 57-bus test systems. The simulation
results are compared with the results obtained using Improved Genetic Algorithm (IGA).
II. STATIC VOLTAGE STABILITY ANALYSIS

The static voltage stability analysis involves determination of an index known as voltage
collapse proximity indicator. This index is an approximate measure of closeness of the system to
voltage collapse. There are various methods of determining the voltage collapse proximity
indicator. One such method is L-index method proposed in [20]. It is based on load flow
analysis. Its value ranges from 0 (no load condition) to 1 (voltage collapse). The bus with the

highest L index value will be the most vulnerable bus in the system. The L-index calculation for
a power system is briefly discussed below:
Consider a N-bus system in which there are Ng generators. The relationship between
voltage and current can be expressed by the following expression:

I G YGG
I Y
L LG

YGL
YLL

VG
V
L

(1)

where, IG, IL and VG, VL represent currents and voltages at the generator buses and load buses.
Rearranging the above equation we get,

VL Z LL FLG
I K

G GL YGG

I L
V
G

(2)

where

FLG YLL YLG

1

(3)

The L-index of the jth node is given by the expression,

Ng

L j 1 F ji
i 1

Vi
( ji i j )
Vj

where
Vi

Voltage magnitude of ith generator.

Vj

Voltage magnitude of jth generator.

ji Phase angle of the term Fji.

(4)

Voltage phase angle of ith generator unit.

j Voltage phase angle of jth generator unit.

Ng Number of generating units.
The values of Fji are obtained from the matrix FLG. The L- indices for a given load condition are
computed for all the load buses and the maximum of the L- indices (Lmax) gives the proximity of
the system to voltage collapse.
The L-index has an advantage of indicating voltage instability proximity of current
operating point without calculation of the information about the maximum loading point.
III.PROBLEM FORMULATION

The optimal power flow problem is formulated as an optimization problem in which a specific
objective function is minimized while satisfying a number of equality and inequality constraints.
The objective of the OPF problem considered here is the minimization of fuel cost in the normal
state and the minimization of the voltage stability index, Lmax in the contingency state. The two
objective functions are obtained by considering a weighting factor w. Power flow equations
are the equality constraints of the problem and the inequality constraints include the limits on
real and reactive power generation, bus voltage magnitude, reactive power generation of var
sources, transformer tap position and line flow and ramp limits.
2.1 Objective function:
The main objective considered in this paper can be expressed as follows:

Ng

Minimize FT =

i 1

Fi + w Lmax, c

(5)

Where Fi = aiPgi2+biPgi+ci $/hr

2.2 System constraints
(i) Load flow constraints:
NB

Pi Vi Vj (GijCosij Bij Sinij ) 0, i 1,2,.......NB1

(6)

j 1

NB

Qi Vi Vj (Gij Sinij BijCosij ) 0, i 1,2,......NPQ

(7)

j 1

(ii) Voltage constraint:

Vimin < Vi < Vimax

i NB

(8)

(iii) Generator reactive power generation limit:

Qgimin < Qgi Qgimax

i Ng

(9)

(iv) Reactive power generation limit of capacitor banks

QCi min < QCi QCi max

i NC

(10)

(v) Transformer tap setting limit:

tkmin < tk < tkmax

k NT

(11)

(vi) Transmission line flow limit

l Nl

Sl < Slmax

(12)

vii) Ramp rate constraints

dP
|PGik- PGi0| TK Gi
dt

max

i NG

(13)

In Equation (5), the first term represents the operating cost of the thermal generating
units. The second term guarantees that the voltage security of the system is improved, and its
effect on the optimization is controlled depending on the weighting factor, w. Depending on the
objectives of the power system operator, emphasis can be placed on one of the components. A
reasonable initial value of weighting factor, w is first selected depending on the OPF problem.
The value of weighting factor is then increased so that stability takes precedence over cost and
hence the system becomes more stable with higher operating costs.
The equality constraints given by Equations (6) and (7) are satisfied by running the power
flow program. The active power generation (Pgi), generator terminal bus voltages (Vgi),
transformer tap settings (tk) and reactive power generation of capacitor bank (Qci) are the
optimization variables and they are self-restricted by the optimization algorithm. The limit on
active power generation at the slack bus (Pgs), load bus voltages (Vload), reactive power
generation (Qgi), line flow (Sl) and ramp rates are satisfied through penalty function approach.
IV. PROPOSED ADAPTIVE TUNED DIFFERENTIAL EVOLUTION
Differential Evolution [21] is a population-based stochastic search algorithm that works in the
general framework of evolutionary algorithms. The optimization variables

are represented

as floating point numbers in the DE population. It starts to explore the search space by randomly
choosing the initial candidate solutions within the boundary. Differential Evolution creates new
off springs by generating a noisy replica of each individual of the population. The individual that
performs better from the parent vector (target) and replica (trial vector) advances to the next
generation. This optimization process is carried out with three basic operations namely,
mutation, crossover and selection. In the proposed work, DE/rand/1/bin strategy with self tuned
parameters is used. Here rand denotes randomly selected vector to be perturbed, 1 denotes the
number of difference vectors considered for perturbation and bin stands for binomial type of

crossover operator. This strategy remains the most competitive scheme based on accuracy and
robustness of results in the OPF problem. The details of these operators are given below:
a. Initialization of parameter vectors:
DE begins with a randomly initiated population of NP real parameter vectors known as
genomes/chromosome which forms a candidate solution to multidimensional optimization
problem expressed as:
Xi,g= x1,i , g , x 2,i , g , x3,i , g ,....x D ,i , g
Where g is the generation number and D is the problems dimension. For each parameter of the
problem, there will be minimum and maximum value within which the parameter should be
restricted. Hence the jth component of ith vector is initialized as follows:
x j ,i , 0 x j , min rand i , j [0,1].x j , max x j ,min

(14)

Where rand i , j [0,1] is a uniformly distributed random number lying between 0 and 1.
b. Mutation with Difference vectors:
After the population is initialized, the mutation operator is in charge of introducing new
parameters into the population. The mutation operator creates mutant vectors by perturbing a
randomly selected vector (Xr1) with the difference of two other randomly selected vectors (Xr2
and Xr3). All of these vectors must be different from each other, requiring the population to be of
at least four individuals to satisfy this condition. To control the perturbation and improve
convergence, the difference vector is scaled by a user defined constant in the range between 0
and 1.2.
Vi ,G X r1,G F ( X r 2,G X r 3,G )

(15)

where F is scaling constant.

In this work, DERANDSF (DE with Random Scale Factor) is used in which the scaled
parameter F is varied in a random manner in the range (0.5,1) by using the relation:
F=0.5 * [1+rand (0,1)]

(16)

Where rand(0,1) is a uniformly distributed random number within the range [0,1]. This allows
for the stochastic variations in the amplification of the difference vector and thus helps retain
population diversity as the search progresses. The difference vector based mutation is believed
to be the strength of DE because of the automatic adaptation in improving the convergence of the
algorithm which comes from the idea of difference based recombination operator ie., Blend
crossover operator (BLX)[22,23].
c.Crossover:
The crossover operator creates the trial vectors which are used in the selection process. A trial
vector is a combination of mutant vector and a parent vector based on different distributions like
uniform distribution, binomial distribution, exponential distribution is generated in the range
[0,1] and compared against a user defined constant referred to as the crossover constant. In this
work, binomial crossover is performed on each of the D variables. If the value of the random
number is less or equal to the value of the crossover constant, the parameter will come from the
mutant vector, otherwise the parameter comes from the parent vector. The crossover operation
maintains diversity in the population preventing local minima convergence. The crossover
constant must be in the range from 0 to 1. If the value of crossover constant is one then the trial
vector will be composed of entirely mutant vector parameters. If the value of crossover constant
is zero then the trial vector will be composed of entirely parent vector. Trial vector gets at least
one parameter from the mutant vector even if the crossover constant is set to zero.

Vi, j (G) if rand(0,1) CR or j q

Ui, j (G)
Xi, j (G) otherwise

(17)

Where q is randomly chosen index in the D dimensional space.

CR is crossover constant
Xi,j(G) is parent vector
V i,j(G) is mutant vector
d.Selection
To keep the population size constant over subsequent generations, the selection process
determines which one of the target vector and trial vector will survive in the next generation and
is outlined as follows:
X i (G 1) U i (G ) if f (U i (G )) f ( X i (G )),
X i (G ) if f ( X i (G )) f (U i (G ))

(18)

Where f(X) is the objective function to be minimized. So if the new trial vector yields a better
value of the fitness function, it replaces its target in the next generation; otherwise the target
vector is retained in the population. This process is continued until the convergence criterion is
satisfied. The termination condition is satisfied when the best fitness of the population does not
change appreciably over successive iterations. This is pictorially represented in Figure 1. The
three main control parameters of DE are : mutation scale factor F,crossover constant CR and the
population size NP. The parameter CR controls how many parameters in expectation are changed
in a population member. The value of F lies in the range of 0.4 and 1.

Initialize population

Run power flows

and evaluate fitness

Select parents for

reproduction

Apply differential
crossover and
mutation

Evaluate fitness of
chromosomes

converged
No
yes
Stop

Figure 1. Flowchart of DE based algorithm

IV.DIFFERENTIAL EVOLUTION IMPLEMENTATION

For the VSC-OPF problem under consideration, generator active power generation (Pgi),
generator terminal voltages ( Vgi ), reactive power generation of VAR sources (Qci) and the
transformer tap positions (tk) are the optimization variables. Real power settings of generators
and generator bus voltages are represented as floating point numbers, whereas the transformer
tap position and the reactive power generation of VAR sources are represented as integers. The
transformer tap setting with tapping ranges of 10% and a tapping step of 0.025 p.u is
represented from the alphabet (0,1,8) and the VAR sources with limits of 1 and 5 p.u and step
size of 1 p.u is represented from the alphabet (0,1,...5). With this representation, a typical
chromosome structure of the OPF problem will look like the following:
97.5 100.8 ... 250.70 0.981 0.970 1.05

Pg2

Pg3

Pgn

Vg1

Vg2

Vgn

t1

1 ... 8

t2 tn

Qc1

1 ..

Qc2

Qcn

Differential Evolution searches for the optimal solution by maximizing a given fitness
function, and therefore an evaluation function which provides a measure of the quality of the
problem solution must be provided. In the optimal power flow problem under consideration, the
objective is to minimize the objective function which comprises fuel cost and Lmax satisfying the
constraints (5-13). The evaluation for the proposed algorithm after the inclusion of penalty
function are as follows:
N PQ

Ng

Nl

Ng

j1

j1

j1

j1

Min f = fc +Ks. SP Kv VPj Kq QPj + Kl LPj +Kr ramp gi

(21)

Here, SP,VPj ,QPj, LPj, rampgi are the penalty terms for the reference bus generator active power
limit violation, load bus voltage limit violation, reactive power generation limit violation, line
flow limit violation, ramp limits and Ks,Kv,Kq,Kl,Kr are the penalty factors. The success of the
penalty function approach lies in the proper choice of these penalty parameters.

V. SIMULATION RESULTS

The proposed Adaptive Differential Evolution algorithm has been applied to voltage stability
constrained optimal power flow problem in IEEE 30 bus system and IEEE 57 bus test systems.
The real and reactive loads are scaled up to predetermined weighting factors to analyze the
system under stressed condition. Real power settings of generators, generator bus voltages,
switchable VAR compensators and transformer tap settings are considered as control variables
for optimal power flow problem. The program was written in MATLAB 7.1 and executed on a
PC with 2.4GHz Intel Pentium IV processor. The results of the simulation are presented below:
IEEE 30-Bus System:
The IEEE 30-bus system has 6 generators, 24 load buses and 41 transmission lines of which four
branches (6-9), (6-10), (4-12) and (28-27) are with the tap changing transformer. Buses
10,12,15,17,20,21,23,24 and 29 are identified for reactive power injection based on the
maximum value of L-indices in the load buses. The detailed data of the system are taken from
[24]. The details of ramp rates of generating units are given in Appendix 1. The lower and upper
bounds of voltage magnitudes are 0.95 and 1.05 pu respectively. The total active and reactive
powers of the system load are 283.4 MW and 126.2 MVAR respectively. The proposed DE

based algorithm was tested with different parameter settings and the best results are obtained
with the following settings:
Population size

: 30

Crossover Rate

: 0.4

Scaled Parameter

: tuned random value

Number of generations

: 150

The proposed DE approach took 185 secs to reach the optimal solution. The optimal values of
the control variables from the proposed algorithm are given in Table 1. The algorithm reached a
minimum cost of 801.93 $/hr and the Lmax of 0.3343. For comparison the results obtained are
compared with the IGA[19] and are given in Table 1. Also, it is clear that the results of DE are
better in minimizing the cost by 16% and improving voltage profile by 9% than improved GA
which took 196 secs to reach the optimal solution. From this table, it is evident that the proposed
DE algorithm is more effective in reaching the optimal solution than the other conventional and
evolutionary approaches. The comparison of fuel cost minimization as objective between the
proposed approach and the other algorithms are presented in Table 2. This shows the
effectiveness of the proposed approach in solving the OPF problem.
Then the system is stressed to 125% loading condition in addition to the outage of critical
line. From the contingency analysis, line outage 28-27 is the most severe contingency with Lmax
value of 0.4165. The proposed algorithm was applied with minimization of contingency state
Lmax as the objective function along with minimum value of control variable adjustments as
additional constraints in the VSC-OPF problem. The ramp rate details of IEEE 30 bus system are
given in Appendix A1. The result of the VSC-OPF problem is given in Table 3.

The fuel cost which is to be increased to 822.53 $/hr due to the reduction in Lmax value is
compensated by including ramp rate constraint in OPF problem and hence the proposed DE
results in 804.1639 $/hr in contingency state which is just 2.23 $/hr greater than the base case
value because of less generator rescheduling of 12.65 MW from the base case value. In addition,
the proposed DE with ramp rate constraints reschedules the generator with lower operating cost
than the ramp rate constrained GA. From Table 4, it is clear that the proposed DE reaches Lmax
value of 0.3203, and the fuel cost is reduced by 18.4% which is less compared to the GA based
VSC-OPF problem and the minimum voltage of the system has increased from 0.8863 to
0.9608, an improvement of about 7.5%. It shows that the proposed algorithm has reduced the
generation cost to obtain more economical corrective control action. The voltage profile of the
system before and after the optimization algorithm with power generation re-dispatch at the load
buses for the critical contingency 28-27 are displayed in Figure 3. Improvement in the voltage
profile of the system with the proposed VSC-OPF algorithm is evident from this diagram.
Further, before the application of the algorithm voltage violations were present in a few buses
but they are corrected after the application of the proposed algorithm. Improvements in voltage
stability and voltage profile have been achieved for the other contingency cases also. This shows
the effectiveness of the proposed algorithm for voltage security enhancement along with
reduction in generation cost.

Table 1 Result of OPF-base case

Improved GA

Adaptive tuned DE

Control

Variable

Variable

Variables

Setting

Control
variables

P1

175.2542

P1

177.6089

P2

48.7936

P2

47.524

P5

21.1651

P5

21.82

P8

22.5074

P8

21.125

P11

12.4202

P11

12.414

P13

12.5815

P13

12.292

V1

1.0497

V1

1.0496

V2

1.0327

V2

1.0356

V5

0.9966

V5

1.0114

V8

1.0077

V8

1.0181

V11

1.0765

V11

1.0943

V13

1.0955

V13

1.0982

T11

0.95

T11

0.975

T12

0.95

T12

0.9

T15

0.975

T15

0.95

T36

0.925

T36

0.925

Qc10

Qc10

Qc12

Qc12

Qc15

Qc15

Qc17

Qc17

setting

Qc20

Qc20

Qc21

Qc21

Qc23

Qc23

Qc2Qc29

Qc24

Qc29

Cost ($/hr)

802.0956

801.93

Loss
(MW)

9.1309

9.0407

Lmax

0.3431

0.3343

Table 2. Comparison of Fuel cost

Method

Minimum cost

Gradient approach [2]

802.43 $/hr

Hybrid Evolutionary
programming [9]

802.62 $/hr

Improved Evolutionary
programming [24]

802.465 $/hr

Improved Genetic algorithm

802.19 $/hr

[19]
Proposed method

801.93 $/hr

Table 3. Results of Optimization under Contingency (28-27) with generation rescheduled

VSCOPF in IEEE 30-Bus System
Control
Variables

GA

DE

Real power
settings

172.864

168.9232,

50.072

50.5053,

23.939

22.3471

23.47

,23.2044,

11.096

14.3455,

12.144

13.3883

1.0475

1.0374,

1.0373

1.0204,

0.9973

1.0190,

1.0091

1.0224,

1.0856

1.0959,

1.0998

1.0996

0.9;

0.975,

0.9;

0.9,

0.9;

1,

1.1;

0.975

Voltage
Magnitudes

Transformer
Tap settings

806.4

804.1639

Change in real
power
settings(MW)

13.2219

12.65

Lmax

0.3320

0.3203

VAR source
Installments

Cost ($/hr)

Table 4 Comparison between GA and DE for line outage (28-27)

Line outage 28-27

No
ramp
rate

Ramp rate

Ramp rate

constrained GA

constrained DE

Lmax

0.4165

0.3320

0.3203

Cost ($/hr)

822.53

806.4

804.13

Vmin

0.8863

0.9502

0.9608

(125% loaded
condition)

Figure 3 Voltage profile improvement for line outage 28-27

IEEE 57 bus test systems:
The IEEE 57-bus system was chosen as the second test system to demonstrate the methods
usefulness on a large system. The details of the IEEE 57-bus system are given below:
No of generators

:4

No of synchronous condensers

:3

No of load buses

: 50

No of transmission lines

: 80

No of tap changing transformers

: 16

The base load of the system is 1272 MW and 298 MVAR. The optimal values of the control
variables along with the minimum fuel cost obtained under base load condition are given in
Table 5. Corresponding to this control variable setting it was found that there are no limit
violations in any of the state variables. The proposed DE algorithm with the inclusion of ramp
rate constraint has brought the total cost to 4.2133x104 $/hr. From Table 5, it is clear that
Proposed DE gives better results by making minimum amount of control variable adjustments as

additional constraint than DE with no generation rescheduling. This shows the powerfulness of
the proposed algorithm to obtain the optimal solution in the OPF problem.
To analyze the system under disturbance condition, contingency analysis was conducted
at 1.25 times the base load condition. From the contingency analysis, line outage 46-47 is found
to be the most severe case with the Lmax value of 0.5068. From the weak bus ranking, buses
30,32,31,33 and 34 were selected for reactive power injection. In contingency state, the voltage
stability indicator, Lmax and ramp rates of generating units are incorporated into the OPF problem
as additional constraints and the proposed DE-based approach was applied to solve the VSCOPF problem. As in the previous case, generator real power settings, generator voltages, shunt
capacitors and OLTC were used as the control variables for improving the voltage security. The
optimal control variable setting after the application of the algorithm for the critical contingency
is summarized in Table 6. From this table, it is found that the value of Lmax decreases along with
generation cost after the application of the proposed DE algorithm than the GA approach. The
voltage profile of the system before and after the application of the algorithm under contingency
46-47 are displayed in Figure 4. Improvement in voltage profile of the system after the
application of the algorithm is evident from this result. The minimum voltage of the system has
been increased by 22% by proposed DE. Hence the improvement in voltage profile of the system
after the application of the proposed algorithm is evident from this result.

Table 5 Results of DE-OPF including minimum control variable adjustment

for IEEE 57 bus test system
DE with ramp rate constraints

DE without ramp
rate constraint

Control

Variable

Variable setting

Variables

setting

P1

157.146

163.0514

P2

55.7916

52.4722

P3

48.1226

48.2904,

P6

17.6880

23.4502

P8

531.641

519.7855

P9

93.293

96.2472

P12

370.262

379.1794

V1

1.0462

1.0565

V2

0.9769

0.9568

V3

1.0242

1.0546

V6

1.0272

1.0435

V8

0.9709

0.9913

V9

0.9734

1.0599

V12

1.0251

1.06

Tap
settings

1.1,0.975,1.025,1.075,

1.05,0.9,1,0.975,

0.9,0.9,1.05,0.925,0.9

1,1,1.1,1.05,0.975

0.95,1.025,1.075,1

0.9,1.05,0.9,1.05,

1.1,1.075,1.025

1.075,0.925,1.1

Var source
installments

0,0,1,2,0

4,5,0,0,1

Cost($/hr)

4.2133x 104

Clmax

0.4748

6.283 x 104

0.5017

Figure 4. Voltage Profile Improvement in IEEE 57 bus test system

Table 6 Results of VSC-OPF for line outage 46-47

Line outage
46-47 (125%

No ramp
rate
constraints

Ramp rate
constrained

Ramp rate
constrained

GA

DE

0.5068

0.5017

0.4748

loaded
condition)
Lmax

8.4799 *104

Cost ($/hr)
Change
real

in

375

4.2506* 104
54

4.2133* 104
38.8759

power

setting (MW)

VI. CONCLUSION

An efficient Differential Evolution algorithm with self tuned parameter has been proposed to
solve voltage stability constrained optimal power flow problem. The weak buses in the system
were selected for reactive power injection. A multiobjective formulation of OPF problem has
been developed in which candidate solutions are selected to reduce the operating costs of thermal
generating units and contingency state Lmax while improving voltage security of the system. In
addition, minimum amount of control variable adjustments in real power settings of generators in
the contingency state were included as additional constraints in the VSC-OPF problem. The
simulation result of IEEE 30 bus system and IEEE 57 bus test system using the proposed DE
approach are compared with Improved GA and is clear that the proposed algorithm improves

voltage stability of the system along with reduction in operating cost by including generation
rescheduling in the contingency state.
VII. REFERENCES
[1] H.W. Dommel and W.F. Tinney, Optimal power flow solutions,

IEEE Transactions on

Power Apparatus and Systems, vol. PAS-87, No 10, Oct 1968, pp 1866-1876.
[2] O.Alsac, B.Scott, Optimal load flow with steady state security, IEEE Trans on power
systems, PAS 93, (3), 1974, pp 745- 751.
[3] K.Y.Lee, Y.M.Park and J.L.Ortiz,

Optimal real and reactive power dispatch,

Electric Power Systems Research, vol 7 1984, pp 201-212.

[4] D.I.Sun et.al, Optimal power flow by Newton approach, IEEE Transactions on PAS,
Vol 103,No 10, 1984, pp 2864-2880.
[5] B.Stott and E.Hobson, Power system security control calculations using

linear

Programming, IEEE Transactions on Power Apparatus and Systems, PAS-97, 1978, pp

1713-1931.
[6] M.K.Mangoli and K.Y.Lee, Optimal real and reactive power control using

linear

Programming, Electric Power Systems Research, vol 26, 1993, pp 1-10.

[7] Florin Capitanescu,M.Glavic,D.Ernst,L.Wehenkel,Interior point based algorithms for the
solution of optimal power flow problems,Electric Power Systems Research, Vol 77,Issue 56,April 2007, pp 508-517.
[8] A.G.Bakirtzis,P.N.Biskas,C.E.Zoumas,V.Petridis Optimal power flow by Enhanced
genetic algorithm, IEEE

Transactions on Power Systems, vol 17, No 2, 2002, pp 229-235.

[9] J.Yuryevich and K.P.Wang, Evolutionary Programming based optimal power flow
algorithms IEEE Transactions on Power Systems, vol 14, No 4, 1999, pp 1245-1250.
[10] Yog Raj Sood, Evolutionary Programming based optimal power flow and its validation for
deregulated power system analysis,International Journal of Electrical Power and Energy
Systems, vol 29, Issue 1, Jan 2007,pp 65-75.
[11] F.Capitanescu,J.L.Martinez Ramos, P.Panciatici,D.Kirschen, A.Marano Marcolini, State-ofthe-art,challenges and future trends in security constrained optimal power flow, Electric
Power Systems Research, 81(2011),pp 1731-1741.
[12] Mithun M.Bhaskar, Muthyala Srinivas and Sydulu Maheswarapu, Security Constraint
Optimal Power Flow (SCOPF)-A comprehensive survey, Trans on Power system, protection
and distribution, Global journal of technology and optimization, vol 2, 2001, pp 11-20.
[13] J.Kubokawa, H.Sasaki,S.Ahmed and G.Strbac, Application of optimal power flow for
voltage stability problem, Proc. International conference on Electric Power Engineering,
Budapest, May 1999, pp 134-141.
[14] C.Canizares, W.Rosehart, A.Berrizi and C.Bovo, Comparison of voltage security
constrained optimal power flow techniques, Proc. IEEE-PES, Summer meeting, Vol 4, No
2, Vancouver, BC.,July 2001, pp 2115-2120.
[15]F.Milano, C.A.Canizares and M.Invernizzi, Voltage stability constrained OPF market
models considering N-1 contingency criteria, Electric Power System Research, Vol 74, No 1,
March 2005, pp 27-36.
[16]A.Monticelli,M.V.F.Pereira,S.Granville,Security contrained optimal power flow with post

contingency corrective rescheduling, IEEE Transactions on power systems,Vol 2, Issue 1,Feb

1987,pp 175-180.
[17]Capitanescu.F and T.V.Cutsem, Preventive control of voltage security margins: a multi
contingency sensitivity based approach, IEEE Transactions in Power systems, 2002, 17(2),
pp 358-364.
[18] G.Y. Wu, C.Y. Chung, K.P. Wong and C.W. Yu, voltage stability constrained optimal
dispatch in deregulated power systems, IET Gener,Trans and Distrib, 2007,1, (5),pp761768.
[19] D.Devaraj and J.Preetha Roselyn, Improved genetic algorithm for voltage security
constrained optimal power flow problem, International Journal of Energy Technology and
Policy, vol5, no 4, 2007, pp 475-488.
[20]P.Kessel and H.Glavitsch, Estimating the voltage stability of power systems, IEEE
Transactions on Power delivery, vol.1, No.3, 1986, pp. 346-354.
[21]Swagatam Das, P.N.Suganthan, Differential Evolution: A survey of the state-of-the-Art,
IEEE Transactions on Evolutionary Computation, Vol 15, No 1, Feb 2011,pp 4-31.
[22] L.J.Eshelman and J.D.Schaffer, Real - coded genetic algorithms and

interval

schemata, in D.Whitley Edition, 1993, pp 187-202.

[23] D.Devaraj, B.Yegnanarayana, Genetic algorithm based optimal power flow for security
enhancement, IEE Proc Gener.Transm.Distrib, Vol 152, Issue 6, 2005, pp899-905.
[24] W. Ongsakul and T. Tantimaporn, Optimal power flow by improved evolutionary
programming, Electric Power Components and Systems, Vol. 34, 2006, pp. 79-95.

VII.

APPENDIX A1

Table A1. Generating units co-efficients with ramp rate limits in IEEE 30 bus system
Generator Pgi min
Ramp
Ramp
Pgi max ai
bi
ci
up limit down
no
(MW/30 limit
min)
(MW/30
mins)
1

50

200

0.001

0.092

14.5

15

20

20

80

0.0004

0.025

22

10

15

15

50

0.0006

0.075

23

10

10

35

0.0002

0.1

13.5

11

10

30

0.0013

0.12

11.5

13

12

40

0.0004

0.084

12.5

10

Table A2. Generating units co-efficients with ramp rate limits in IEEE 57 bus system
Generator Pgi min Pgi max ai
bi
ci
Ramp up Ramp
no
limit
down
(MW/30 limit
min)
(MW/30
mins)
1

575.88

0.0775795

20

50

80

100

0.01

40

10

15

140

0.25

20

10

15

100

0.01

40

10

15

550

0.02222222 20

50

80

100

0.01

40

10

15

12

410

0.0322581

20

30

60