Electrical Power and Energy Systems 30 (2008) 435441

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

Differential evolutionary algorithm for optimal reactive power dispatch

M. Varadarajan, K.S. Swarup *
Department of Electrical Engineering, Indian Institute of Technology, Madras, Chennai 600 036, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents differential evolutionary algorithm for optimal dispatch for reactive power and volt-
Received 27 June 2007 age control in power system operation studies. The problem is formulated as a mixed integer, nonlinear
Received in revised form 5 March 2008 optimization problem taking into account both continuous and discrete control variables. The optimal
Accepted 9 March 2008
setting of control variables such as generator voltages, tap positions of tap changing transformers and
the number of shunt reactive compensation devices to be switched for real power loss minimization in
the transmission system are determined. In the proposed method, the inequality operational constraints
Keywords:
were handled by penalty parameterless approach. This helps in avoiding the time-consuming trial and
Evolutionary computation
Differential evolution
error process for xing the penalty parameter and makes the process system independent. The algorithm
Optimal power ow was tested on standard IEEE 14,30,57 and 118-Bus systems and the results compared with conventional
Loss minimization method.
Penalty function 2008 Elsevier Ltd. All rights reserved.

1. Introduction A number of mathematical programming based techniques

Global optimization of a non-continuous, nonlinear function,
arising from large-scale complex engineering problems, which
may have a large number of local minima and maxima, is quite
challenging. A number of deterministic approaches based on
branch and bound and real algebraic geometry are found to be suc-
cessful in solving these problems to some extend. Of late, stochas-
tic and heuristic optimization techniques such as evolutionary
algorithms (EA) have emerged as efcient tools for global optimi-
zation and have been applied to a number of engineering problems
in diverse elds. For the secure and economic operation of large-
scale power systems, a variety of optimization problems have to
be solved. The optimal power ow (OPF) problem, which was
introduced in 1960s by Carpentier, [1] is an important and power-
ful tool for power system operation and planning. Reactive power
optimization is a sub-problem of OPF calculation, which deter-
mines all the controllable variables, such as tap ratio of transform-
ers, output of shunt capacitors/reactors, reactive power output of
generators and static reactive power compensators etc., and mini-
mizes transmission losses or other appropriate objective functions,
while satisfying a given set of physical and operational constraints.
Since transformer tap ratios and outputs of shunt capacitor/reac-
tors have a discrete nature, while reactive power outputs genera-
tors, bus voltage magnitudes and angles are, on the other hand,
continuous variables, the reactive power optimization problem is
formulated as mixed-integer, nonlinear problem.
* Corresponding author. Tel.: +91 44 2257 4440; fax: +91 44 2257 4401.
E-mail address: swarup@ee.iitm.ac.in (K.S. Swarup).
have been proposed to solve the OPF problem. For decades, con-
ventional gradient-based optimization algorithms have been used
for the solving optimal reactive power dispatch problem [2,3].
The gradient and Newton methods suffer from the difculty in
handling inequality constraints. Linear programming requires
objective function and constraints have linear relationship, which
may lead to loss of accuracy. Conventional methods are not ef-
cient in handling problems with discrete variables. The combinato-
rial-search approaches, branch-and-bound and cutting plane
algorithms, which are usually used to solve the mixed integer pro-
gramming model, are non-polynomial and all suffer from the
problem of curse of dimensionality making them unsuitable for
large-scale OPF problems.
Wu and Ma [4] applied Evolutionary Programming (EP) for
global optimization problems of large-scale power systems to
achieve optimal reactive power dispatch and voltage control of
power systems. Lai and Ma [5] showed that in optimization of
non-continuous and non-smooth function, EP is much better
than nonlinear programming and has applied it for reactive
power planning. Lee et al. [6] solved the reactive power opera-
tional and investment-planning problem by using a Simple
Genetic Algorithm (SGA) combined with the successive linear
programming method. The Benders cut is constructed during
the SGA procedure to enhance the robustness and reliability of
the algorithm. Chebbo and Irving [7] proposed a linear program-
ming based conventional approach for combined active and reac-
tive power dispatch. Yoshida et al. [8] proposed a Particle Swarm
Optimization (PSO) for reactive power and voltage control con-
sidering voltage security assessment. Zhao et al. [9] proposed a

0142-0615/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijepes.2008.03.003
436 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441

solution to the reactive power dispatch problem with a PSO as capacitor and reactor banks, i.e., u P g ; V g ; t; Q c T . Of the control
approach based on multi-agent systems. variable mentioned Pg and V g are continuous variables, while tap
Differential evolutionary algorithm (DEA) is a technically sim- ratio, t, of tap changing transformers and reactive power output
ple; population based evolutionary algorithm (EA), which is highly of compensation devices, Q c , are discrete in nature. Loss minimiza-
efcient in constrained parameter optimization problems [10]. tion is usually required when cost minimization is the main goal
DEA employs a greedy selection process with implicit elitist fea- with generator active power generation as the control variable.
tures. It has demonstrated its robustness and effectiveness in a When all control variables are utilized in a cost minimization, a sub-
variety of applications, such as neural network learning and in- sequent loss minimization will not yield further improvements.
nite impulse response lter design [11,12]. It presents no difculty Therefore in reactive power dispatch problem, such as loss minimi-
in solving mixed integer problems [13] and hence is highly suitable zation, active power generation of all generators, except slack gen-
for reactive power optimization where the generator voltage is a erator, is xed during the optimization procedure.
real valued parameter while tap position and the number of shunt
devices to be switched is integer parameters. DE differs from other 3. Problem formulation
EAs in the mutation and recombination phase. Unlike stochastic
techniques such as Genetic Algorithms (GA) and Evolutionary The solution of the optimal reactive power dispatch problem in-
Strategies (ES), where perturbation occurs in accordance with a volves the optimization of the nonlinear objective function with
random quantity, DE uses weighted differences between solution nonlinear system constraints.
vectors to perturb the population. Authors in [1417] used differ-
ential evolution to solve problems in power systems. 3.1. Objective function
Method of constraint handling is extremely important in power
system optimization problems. In all the previous works reported The objective function here is to minimize the active power loss
in literature, inequality constraints were handled by use of a pen- (Ploss ) in the transmission system. Network losses either for the
alty function approach, i.e., the constraint violation is multiplied by whole network or for certain sections are non-separable functions
a penalty coefcient or parameter and added to the objective func- of dependent and independent variables. It is given as
tion. Deb [18] proposed a penalty parameterless scheme to over-
Nl
X
come the difculty of choosing penalty coefcients for GA based Ploss g k tk V i 2 V 2j  2tk V i V j cos hij  2
constrained optimization problems. It is important to realize that k1
such a penalty parameterless strategy is only applicable to popula-
where, Nl is the number of transmission lines; g k is conductance of
tion based approach. This is because it requires the population to
branch k between buses i and j; t k the tap ratio of transformer k; V i
be divided into two sets: feasible and infeasible sets. The tness
is the voltage magnitude at bus i; hij the voltage angle difference be-
function depends on the feasible and infeasible population mem-
tween buses i and j.
bers. Since in a point-by-point optimization approach, there is only
one member in each iteration, such a penalty parameterless
3.2. Constraints
scheme cannot be applied. Although a penalty term is added to
the objective function to penalize infeasible solutions, the method
The minimization of the above function is subjected to a num-
differs from the way the penalty term is dened in conventional
ber of equality and inequality constraints. The equality constraints
methods and in earlier evolutionary algorithm implementations.
are the power ow equations given by
In this paper an optimal reactive power dispatch using differential
evolutionary algorithm with an efcient penalty parameterless Nb
X
scheme of constraint handling is employed. A performance com- Pgi  P di  V i V j Gij cos hij Bij sin hij 0 for i 1; . . . ; N PV N PQ
j1
parison with conventional interior point technique is provided to
highlight the efciency of the differential evolutionary algorithm 3
based optimization. Nb
X
Q gi  Q di Q ci  V i V j Gij sin hij  Bij cos hij 0 fori 1; . . . ; N PQ
j1
2. Optimal power ow
4
The optimal power ow (OPF) is a static, nonlinear optimization where, Nb , N PV and N PQ are the number of buses, PV buses and PQ
problem, which calculates a set of optimum variables from the net- buses respectively; Gij , Bij are real and imaginary part of i; jth ele-
work state, load data and system parameters. Optimal values are ment of bus admittance matrix; Pgi , Q gi are active and reactive
computed in order to achieve a certain goal such as generation cost power generation at bus i ; Pdi , Q di are active and reactive power de-
or line transmission power loss minimization subject to equality mand at bus i ; Q ci the reactive power compensation source at bus i.
and inequality constraints. The OPF problem can be presented as The inequality constraints on security limits are given by
min f x; u Pmin max
g;slack 6 P g;slack 6 P g;slack 5
s:t gx; u 0 1
V min
Li 6 V Li 6 V max
Li for i 1; . . . ; N PQ 6
hx; u 6 0
Q min
gi 6 Q gi 6 Q max
gi for i 1; . . . ; N g 7
where, f is the objective function that typically includes total gen-
eration cost, losses in transmission system etc. Generally, gx; u jSl j 6 Smax
l for l 1; . . . ; N l 8
represents the load ow equations and hx; u represents transmis- The inequality constraints on control variable limits are given by
sion line limits and other security limits. The vector of dependent
and control variables are denoted by x and u respectively. In gen- V min
gi 6 V gi 6 V max
gi for i 1; . . . ; N PV 9
eral, the dependent vector includes bus voltage angles h, bus voltage tmin
k 6 t k 6 t max
k for k 1; . . . ; N t 10
magnitudes V L and generator reactive power Q g , i.e., x h; V L ; Q g T . Q min 6 Q ci 6 Q max for i 1; . . . ; N c 11
ci ci
The control variable vector consists of real power generation P g ,
generator terminal voltage V g , transformer tap ratio t and reactive where, N g , Nc and Nt are the number of generators, compensator de-
power generation or absorption Q c of compensation devices such vices and transformers; Sl the apparent power ow in line l ; Smax
l
 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441 437

(
the maximum apparent power ow in line l; V gi the voltage magni- U G1 if f U G1 < f X Gi
tude at generator bus i; V Li the voltage magnitude at load bus i; X iG1 i i
15
min
X Gi otherwise
Pg;slack the real power generation at slack bus; V min max
Li , V Li , Q gi ,
max min max min max min max min max
Q gi , t k , t k , V gi , V gi , Q ci , Q ci , Pg;slack , and Pg;slack are mini- All solutions in the population have the same chance of being se-
mum and maximum limits of the corresponding variables lected as parents independent of their tness value. If the parent
respectively. is still better, it is retained in the population, thus incorporating
the feature of elitism.
4. Differential evolutionary algorithm
4.4. Termination criteria
Differential Evolution Algorithm (DEA) is a simple population-
based, stochastic parallel search evolutionary algorithm for global The iterative procedure can be terminated when any of the fol-
optimization and is capable of handling non-differentiable, nonlin- lowing criteria is met, i.e., an acceptable solution has been reached,
ear and multi-modal objective functions. In DEA the population a state with no further improvement in solution is reached, control
consists of real-valued vectors with dimension D that equals the variables has converged to a stable state or a predened number of
number of design parameters. The size of the population is ad- iterations have been completed. In most of the cases, it is not easy
justed by the parameter N p . The initial population is uniformly dis- to test whether the obtained solution is the most acceptable one.
tributed in the search space. Here, in the present work, the iterations are terminated whenever
the results remain constant for a xed number of generations or
xGi;k xkmin rand0; 1  xkmax  xkmin i 2 1; N p ; maximum number of generations is reached, whichever occurs
k 2 1; D 12 rst.

Each variable k in a individual i in the generation G is initialized

within its boundaries xkmin and xkmax . At each generation, two opera-
tors, namely mutation and crossover (recombination), are applied
on each individual, thus producing the new population. Then, a
selection phase takes place, where each individual of the new pop-
ulation is compared to the corresponding individual of the old pop-
ulation, and the best between them is selected as a member of the
population in the next generation. In the following, the evolutionary
operators are briey described.
4.1. Mutation
DE does not use a predened probability density function to
generate perturbing uctuations. It relies upon the population it-
self to perturb the vector parameter. According to the mutation
operator, for each individual, X Gi , i 1; . . . ; N p , at generation G, a
mutant vector, V iG1 vG1 G1 G1
i1 ; vi2 ; . . . ; viD is determined using
the following equation
V G1 X Gr1 sX Gr1  X Gr3
i culty arises here in choosing the appropriate penalty parameter R.
where, s > 0 is a real parameter, called mutation constant, which
controls the amplication of the difference between two individuals
so as to avoid search stagnation and r 1 , r 2 and r 3 are mutually dif-
ferent integers, randomly selected from the set f1; 2; . . . ; N p g.
4.2. Crossover
5. Constraint handling
5.1. With penalty coefcients
A common practice, for handling nonlinear constraints, is to use
a penalty function approach because of its simplicity and ease of
implementation. In most applications of EA to constrained optimi-
zation problems, this penalty function approach has been used. In
this method for handling inequality constraints in minimization
problems, the tness function Fx is dened as the sum of the
objective function f x and a penalty term which depends on the
constraint violation hhxi.
X
n
Fx f x Rj hhj xi2 16
j1
where, denotes the absolute value of the operand, if the operand is
negative and returns a value zero, otherwise. The parameter Rj is
the penalty parameter of the jth inequality constraint. The main dif-
13
The values of these parameters are chosen by trial and error meth-
od. To highlight the inuence of the choice of penalty parameters on
the objective function value, a simulation study is carried out on
IEEE 14-Bus system [8]. The results obtained for three different
cases are listed in Table 1. It can be observed that the real power
loss value is greatly dependent on the choice of penalty parameters.

Following the mutation phase, the crossover (recombination) 5.2. Without penalty coefcients
operator is applied on the population. For each mutant vector
V G1 , a trial vector, U iG1 ui1
G1
; uG1 G1 In penalty parameter-less scheme, employed in this work, the
i i2 ; . . . ; uiD is generated, with
( G1 composite tness function for any f x is given as follows
vj;i if randj 6 CR or j Irand 
uG1
j;i 14 f x if x is feasible
xGj;i if randj > CR and j 6 Irand Fx 17
fmax CVx otherwise
where randj 2 0; 1 and Irand is chosen randomly from the interval
Here, fmax is the objective function value of the worst feasible solu-
1; . . . ; D once for each vector to ensure that at least one vector
tion in the population. In situation where none of the solutions in a
component originates from the mutated vector vi . CR is the DE con-
trol parameter that is called the crossover rate and is a user dened
parameter within range [0,1]. Equation is applied for every vector
Table 1
component i 2 1; . . . ; N p , j 2 1; . . . ; D.
Effect of penalty parameters on the solution

4.3. Selection Parameter Case 1 Case 2 Case 3

R1 1000 2000 3000
To decide whether the vector U iG1 should be a member of the R2 100 200 300
R3 100 100 100
population comprising the next generation, it is compared to the
P loss (MW) 13.42 13.29 13.32
corresponding vector X Gi .
438 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441
population are feasible, fmax is not dened. Hence, such situations
are handled by articially inserting the base case solution into the
population. CVx is the overall constraint violation of solution x.
It is calculated as follows
CVx max0; P g;slack  P max min
g;slack ; P g;slack  P g;slack
N PQ
X
max0; V Li  V max min
Li ; V Li  V Li
i1
Ng
X Nl
X The details of the DE based optimization algorithm are as
max0; Q gi  Q max min
gi ; Q gi  Q gi maxjSl j  Smax
i1 l1
Since all feasible solution have zero constraint violation and all
infeasible solutions are evaluated according to their constraint vio-
lations only, both the objective function value and constraint viola-
tion are not combined in any solution in the population. Thus there
is no need to have any penalty coefcient R for this approach. The
advantages of this scheme as compared to the usual penalty param-
eter based scheme are (i) The tedious process of choosing a suitable
penalty coefcient R can be avoided, the inappropriate choice of
which will affect the nal solution and (ii) there is no need to eval-
uate the objective function value for individuals with constraint
violation, which reduces the computation time.
6. Differential evolution for reactive power dispatch
The implementation of differential evolutionary algorithm for
reactive power dispatch consists mainly of identication objectives
and handling of constraints. The control variables selected for reac-
tive power optimization are the generator voltages except for slack
generator; tap position of tap changing transformers and the num-
ber of shunt compensation banks to be switched. DEA is employed
to nd the best control variable setting starting from a randomly
generated initial population of control variables. At the end of each
generation, the best individuals, based on the tness value, are
stored.
6.1. Treatment of control variables
In its basic form, differential evolutionary algorithm can handle
only continuous variables. However, reactive power source instal-
lation and tap position of tap changing transformers are discrete
variable or integer variable and generator terminal voltages are
continuous variables in the reactive power dispatch problem. Here,
differential evolutionary algorithm has been extended to handle
mixed integer variables. For integer variables the value is rounded
off to the nearest integer value of the variable.
 used for the simulation are summarized in Table 2. Number of indi-
xi for continuous variables
xi 19
bxi c for integer variables
The bxc function gives the nearest integer less than or equal to x.
Different control variables in the optimal reactive power dispatch
are treated as follows: initial generator terminal voltages are gener-
ated randomly between upper and lower limits of the voltage spec-
ication values. The value is then modied in the search procedure,
within the specied limits. Transformer tap position is initially gen-
erated randomly between the minimum and maximum tap posi-
tions. The value is then modied in the search procedure among
existing tap positions. Based on the tap position, the corresponding
tap ratio is calculated as follows
tk t min
k nk  Dtk
where, nk is the number of tap positions and Dt k the step size of tap
ratio. Using this tap ratio, the corresponding admittance of the trans-
former is determined for the load ow calculation. Initial number of
reactive power compensation capacitor/reactor banks at a bus is
randomly generated from 0 to the number of existing equipment
at the bus. This value is also modied in the search procedure, al-
ways limiting it between 0 and the number of existing compensator
banks.
6.2. Algorithm
l follows
18
Step. 1 Generate an initial population randomly within the
control variable bounds.
Step. 2 For each individual in the population, run power ow
algorithm such as Newton Raphson method, to nd
the operating points.
Step. 3 Evaluate the tness of the individuals according to Eqs.
(17) and (18).
Step. 4 Perform differentiation (mutation) and crossover as
described in Sections 4.1 and 4.2 to create offspring
from parents.
Step. 5 Perform Selection as described in Section 4.3. between
parent and offspring. While using the penalty parame-
terless method of constraint handling the following cri-
teria are enforced while selecting the individuals for the
next generation.
 Any feasible solution is preferred to any infeasible solution.
 Among two feasible solutions, the one having better objec-
tive function value is preferred.
 Among two infeasible solutions, the one having smaller con-
straint violation is preferred.
Step. 6 Store the best individual of the current generation.
Step. 7 Repeat steps 2 to 6 till the termination criteria is met
(maximum number of generations).
7. Simulation results
Results obtained by simulation of the algorithm using differen-
tial evolutionary algorithm, done in MATLAB 6.5 environment, are
provided in this section. Simulation is carried out on IEEE 14 [8],
IEEE 30, IEEE 57 and IEEE 118 [19] bus test systems. Floating point
representation is used for encoding the control variables. There are
several variants of DE. For this work, the DE scheme, which can be
classied using notation DE=rand=1=bin strategy, is used. This
strategy is the most popularly used one for solving practical prob-
lems [13]. The differential evolutionary computation parameters
viduals in a population for each test system is decided based on
trial simulation run. The results, which follow, are the best solu-
tions over 30 independent trials.
7.1. Case study of IEEE 14-Bus system
In the modied IEEE 14-Bus system shown in Fig. 1, there are 14
buses, out of which 5 are generator buses. Bus 1 is the slack bus, 2,
3, 6 and 8 are taken as PV generator buses and the rest are PQ load
buses. The network has 20 branches, 17 of which are transmission
lines and 3 are tap- changing transformers. It is assumed that shunt
compensation capacitor is available at buses 9 and 14 for voltage
control. Totally, there are 9 control variables, which consists of 4
20
PV generator voltages, 3 tap changing transformers with 20 steps
of 0.01 p.u. each and 2 shunt compensation capacitor banks with
3 steps of 0.06 p.u. each. The system data and initial operating con-
ditions of the system are given in [8].
 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441 439

Table 2 1.12
Simulation parameters used for differential evolution

DE control parameters 1.1

Population size = 30 Step size s 0:6
Max. no. of generations = 200 Crossover rate (CR) = 0.8
1.08
VG8
VG2
1.06 VG6

VG [p.u]
13
1.04
14 VG3
12
11
1.02
10

1 6 8 1
9

7 0.98
5 0 10 20 30 40 50 60 70
4
Number of Generations

Fig. 3. Convergence of generator voltages for IEEE 14-Bus system.

3
25
Tap Position Number
20
15
Fig. 1. Network diagram of IEEE 14-Bus system.
10 T49
T47 T56
Fig. 2 gives the performance of the optimization technique in 5
terms of Ploss and Q loss . It can be observed that both P loss and Q loss 0
0 10 20 30 40 50 60 70
get reduced over the evolutions and converge to a minimum value.
The iterative procedure is terminated when there is no change in
4
the result for 40 consecutive iterations or when 200 generations
No. of Capacitor Banks Switched
are reached, whichever occurs rst. A good optimization results 3
in convergence of all control variables to a steady state. Fig. 3 Q
shows the variation of the continuous control variable, i.e., gener- 2 C9
Q
C14
ator voltages V g , with respect to the number of generations. All
four-generator voltages settle to a steady state before 30 genera- 1
tions. Fig. 4 shows the variation of the discrete control variables,
0
tap position and capacitor bank switching. It can be observed that 0 10 20 30 40 50 60 70
all discrete control variables also converge well before 30 Number of Generations
generations.
Fig. 4. Convergence of tap position and capacitor bank switching for IEEE 14-Bus
Table 3 gives the details of the control variables before and after system.
optimization. The minimum and maximum limits of load bus volt-

Table 3
Control variable setting and Ploss before and after optimization for IEEE 14-Bus system

Variable Base Case DE PSO IPM

V g2 1.0450 1.0449 1.0463 1.0449
V g3 1.0100 1.0146 1.0165 1.0149
V g6 1.0700 1.1000 1.1000 1.0971
V g8 1.0900 1.1000 1.1000 1.0999
T 47 0.9467 1.0600 0.9400 1.0238
T 49 0.9524 1.0400 0.9300 1.0998
T 56 0.9091 1.1000 0.9700 1.0550
Q c9 0.1800 0.1800 0.1800 0.1798
Q c14 0.1800 0.0600 0.0600 0.0739
P loss (MW) 13.49 13.239 13.327 13.246

Fig. 2. P loss and Q loss variation with generation for IEEE 14-Bus system.
ages are 0.95 and 1.1 p.u. respectively. From the base case value of
13.49 MW, the P loss was reduced to 13.239 MW with differential
evolutionary computation. In order to evaluate the performance
of differential evolutionary computation, the results were com-
pared with popular Particle Swarm Optimization (PSO) [8] and
conventional Interior Point Method (IPM), results of which are also
presented in Table 3. OPF problem is modeled in AMPL [20] and
440 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441

solved using the solver KINTRO [21] assuming all control variables 0.136
to be continuous.
To study the performance of the algorithm for different initial 0.1355
population size 20
population, simulation is carried out with a constant population
population size 30
size of 30 on the test systems. Fig. 5 shows the effect of starting 0.135
population size 40
point on the nal solution for ve trial cases. It can be observed population size 50

Objective Value [p.u]

that even though the trajectory of convergence is different, all tri- 0.1345
als converge to the same nal solution. As in the case of all random
processes, it is useful to do repeated simulations and observe the 0.134
solution statistics. For this particular case study, the simulations
0.1335
were repeated for 30 times from different starting points and the
important statistical details are given in Table 4. It can be seen that
0.133
the standard deviation of 30 trials is very small, guaranteeing a
near best solution for any random trial.(See Table 4).
0.1325
It is essential to study the effect of population size on the opti-
mization procedure. Even though a number of heuristic relations
0.132
are available for nding the population size, there is no hard and 0 10 20 30 40 50 60 70
fast rule, which can be universally adopted. In most cases, the pop- Number of Generations
ulation size is xed by trial and error method. Fig. 6 shows the ef- Fig. 6. Effect of population size on objective value.
fect of population size on the solution, with all four trials having
the same starting point. Except for a population size of 20, which
has delayed convergence, all higher population sizes give more Table 5
or less same speed of convergence. Hence, to achieve a compro- Ploss obtained by various methods for various test systems
mise between fast convergence and reduced computation time, a
IEEE 30-Bus IEEE 57-Bus IEEE 118-Bus
population size of 30 was used for this study.
Base Case 5.660 27.8637 132.450
DE 5.011 25.0475 129.579
7.2. Case study of IEEE 30, 57 and 118-Bus systems PSO 5.092a 25.3047 131.908a
IPM 5.101a 26.0100 132.110a
In this section performance of differential evolutionary algo- a
Taken from reference [22].
rithm for optimal reactive power dispatch was evaluated on IEEE
30, 57 and 118 bus systems with simulation parameters given in
Table 2. The network data for the test systems are given in [19].
As in the case of IEEE 14-Bus system, it was found that all of the Table 6
control variables settled to a steady value when the termination Statistical details for IEEE 30,57 and 118-Bus systems

Compared Item IEEE 30-Bus IEEE 57-Bus IEEE 118-Bus

P loss best (MW) 5.011 25.0475 128.318
P loss worst (MW) 5.022 25.2016 129.579
P loss average (MW) 5.013 25.1112 129.0817
Standard deviation 0.0026 0.049 0.345
Average iteration 66 141 193
Average CPU time (sec) 13.647 35.654 42.1556

criterion is satised. It is assumed that tap changing transformers

Fig. 5. Effect of starting point on objective value P loss .
have 20 discrete steps of 0.01 p.u. each. The available reactive pow-
ers of capacitor banks are within the interval (0-30) MVAr in dis-
crete steps of 1 MVAr. All bus voltages are required to be
maintained within the range of 0.95-1.1 p.u. Table 5 shows the
minimum P loss obtained by various methods namely DE, PSO and
IPM. Differential evolutionary algorithm nds the minimum (opti-
mal) value compared to other methods. To verify the robustness of
the proposed methodology simulation is carried out for 30 trial
runs with different initial population. The important statistical de-
tails are listed in the Table 6. To ensure a near optimum solution
any random trial, the standard deviation should be very low. It
can be seen that DEA is robust.

Table 4 8. Conclusion
Statistical details for IEEE 14-Bus system

Compared Item DE A differential evolutionary computation based optimization

P loss best (MW) 13.2396
method for reactive power dispatch was presented in this section.
P loss worst (MW) 13.2746 It has been observed that differential evolutionary algorithm is a
P loss average (MW) 13.2506 simple but powerful tool for power system optimization problem
Standard deviation 1:6165  e2 with nonlinear objectives and constraints. Optimal control vari-
Average iteration 62
ables for real power loss minimization are determined using the
Average CPU time (sec) 8.172
differential evolutionary computation optimization strategy,
 M. Varadarajan, K.S. Swarup / Electrical Power and Energy Systems 30 (2008) 435441
employing an efcient penalty parameterless method for con-
straint handling. Comparison with conventional techniques has
veried the performance capability of differential evolutionary
algorithm.
References
[1] Dommel HW, Tinney WF. Optimal power ow solutions. IEEE Trans Power App
Syst 1968;PAS-87(10):186676.
[2] Torres GL, Quintana VH. An interior-point method for nonlinear optimal power
ow using rectangular coordinates. IEEE Trans Power Syst 1998;13(4):12118.
[3] da Costa GRM, Costa CEU, de Souza AM. Comparative studies of optimization
methods for the optimal power ow problem. Electric Power Syst Res
2000;56(3):24954.
[4] Wu QH, Ma JT. Power system optimal reactive power dispatch using
evolutionary programming. IEEE Trans Power Syst 1995;10(3):12439.
[5] Lai LL, Ma JT. Application of evolutionary programming to reactive power
planning comparison with nonlinear programming approach. IEEE Trans
Power Syst 1997;12(1):198206.
[6] Lee KY, Bai X, Park YM. Optimization method for reactive power planning by
using a modied simple genetic algorithm. IEEE Trans Power Syst 1995;10(4):
184350.
[7] Chebbo AM, Irving MR. Combined active and reactive dispatch Part 1:
Problem formulation and solution algorithm. IEEE Proc Generat Transmission
Distribut 1995;142(4):393400.
[8] Yoshida H, Fukuyama Y, Kawata K, Takayama S, Nakanishi Y. A particle swarm
optimization for reactive power and voltage control considering voltage
security assessment. IEEE Trans Power Syst 2001;15(4):12329.
[9] Zhao B, Guo CX, Cao YJ. A multiagent-based particle swarm optimization
approach for optimal reactive power dispatch. IEEE Trans Power Syst
2005;20(2):10708.
441
[10] Storn R, Price K. Differential evolution a simple and efcient adaptive scheme
for global optimization over continuous spaces. Technical Report TR-95-012,
ICSI; March 1995.
[11] Storn R, Price K. Differential evolution A simple and efcient heuristic for
global optimization over continuous spaces. J Global Optim 1997;11(4):
34159.
[12] Storn R. System design by constraint adaptation and differential evolution.
IEEE Trans Evol Comput 1999;3(1):2234.
[13] Price K, Storn R, Lampinen J. Differential evolution: a practical approach to
global optimization. Springer; 2005.
[14] Chang CS, Lu LR, Wang F. Application of differential evolution in harmonic
worst-case identication of mass rapid transit power supply system. Int J Syst
Sci 2004;35(1314):7319.
[15] Chiou JP, Chang CF, Su CT. Capacitor placement in large-scale distribution
systems using variable scaling hybrid differential evolution. Int J Electric
Power Energy Syst 2006;28(10):73945.
[16] Liang CH, Chung CY, Wong KP, Duan XZ. Parallel optimal reactive power ow
based on cooperative co-evolutionary differential evolution and power system
decomposition. IEEE Trans Power Syst 2007;22(1):24957.
[17] Chang CF, Chiou JP. Optimal feeder reconguration for loss reduction by robust
searching hybrid differential evolution. Cybernetics and Systems 2007;38(4):
42950.
[18] Deb K. An efcient constraint handling method for genetic algorithms. Comput
Meth Appl Mech Eng 2000;186(24):31138.
[19] System data (2006, December). Available from: <www.ee.washington.edu/
research/pstca>.
[20] Fourer R, Gay DM, Kernighan BW. AMPL: a modeling language for
mathematical programming. Brooks Cole Publishing Company; 2002.
[21] Waltz RA, Todd D. Plantenga TD. KNITRO 5.1 User Manual, Ziena Optimization
Inc.
[22] Vlachogiannis JG, Lee KY. A comparative study of particle swarm optimization
for optimal steady state performance of power Systems. IEEE Trans Power Syst
2006;21(4):171828.