Ain Shams Engineering Journal (2015) xxx, xxx-xxx

Ain Shams University Ain Shams Engineering Journal

www.elsevier.com/locate/asej www.sciencedirect.com

A review of meta-heuristic algorithms for reactive power planning problem

Abdullah M. Shaheen a*, Shimaa R. Speab, Sobhy M. Farragc, Mohammed A. Abido d

a South Delta Electricity Distribution Company (SDEDCo), Ministry of Electricity, Tanta, Egypt b Electrical Engineering Department, Faculty of Engineering, Menoufiya University, Egypt

c Electrical Power Systems, Electrical Engineering Department, Faculty of Engineering, Menoufiya University, Egypt d Electrical Engineering Department, King Fahd University of Petroleum and Minerals, Saudi Arabia

Received 14 June 2015; revised 9 October 2015; accepted 4 December 2015

KEYWORDS

Reactive power planning; Multi-objective optimization;

Arithmetic programming methods;

Meta-heuristic optimization

techniques;

Hybrid techniques

Abstract Reactive power planning (RPP) is generally defined as an optimal allocation of additional reactive power sources that should be installed in the network for a predefined horizon of planning at minimum cost while satisfying equality and inequality constraints. The optimal placements of new VAR sources can be selected according to certain indices related to the objectives to be studied. In this paper, various solution methods for solving the RPP problem are extensively reviewed which are generally categorized into analytical approaches, arithmetic programming approaches, and meta-heuristic optimization techniques. The research focuses on the disparate applications of meta-heuristic algorithms for solving the RPP problem. They are subcategorized into evolution based, and swarm intelligence. Also, a study is performed via the multi-objective formulations of reactive power planning and operations to clarify their merits and demerits. © 2015 Faculty of Engineering, Ain Shams University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Nowadays, reactive power planning (RPP) problem has become one of the most challenging problems in power

Corresponding author. Tel.: +20 1141520206. E-mail addresses: engabdoushaheen@yahoo.com (A.M. Shaheen), shi_spea@yahoo.com (S.R. Spea), sobhy_f@hotmail.com (S.M. Farrag), mabido@kfupm.edu.sa (M.A. Abido). Peer review under responsibility of Ain Shams University.

systems. It has been an important stage of transmission expansion planning (TEP) problem in recent years [1-3]. In addition, reactive power control/dispatch is an important function in the planning process for the future of power systems. It aims to utilize all the reactive power sources efficiently, which are suitably located and sized in the planning process [4-10].

Generally, the various RPP solutions are divided into three groups which are analytical approaches [11-13], arithmetic programming approaches [3,4,11,12-15,16(Ch. 2),17(Ch. 3),18-23], and meta-heuristic optimization techniques. Various Meta-heuristic Optimization Algorithms (MOA) have been applied to the RPP problem such as Genetic Algorithms (GA) [5,24-33], Differential Evolution (DE) [6,17,24,34-42],

http://dx.doi.org/10.1016/j.asej.2015.12.003

2090-4479 © 2015 Faculty of Engineering, Ain Shams University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Harmony Search (HS) [43-45], Seeker Optimization Algorithm (SOA) [46-48], Evolutionary Programming (EP) [49-54], Ant Colony Optimization (ACO) [7,55], Immune Algorithm (IA) [8], Particle Swarm Optimization (PSO) [2,9,16,56-58], Artificial Bee Colony (ABC) [59], Gravitational Search Algorithm (GSA) [60,61], Firefly Algorithm (FA) [62], Teaching Learning Algorithm (TLA) [63], Chemical Reaction Optimization (CRO) [64], Water Cycle Algorithm (WCA) [65], and Differential Search Algorithm (DSA) [66]. Hybrid techniques have been suggested in some researches that make use of advantages of different algorithms simultaneously to improve the quality of solution [5,10,16(Ch. 5),53,55,67-75].

Also, multi-objective formulation of optimization problems for reactive power planning and operation has been treated using the mathematical sum approach [1,11,24,25,28,35-38,5 0,51,53,56,68], weighting functions [27,29,40,43,44,47,69], e-constraint approach [6,18,20,43,76,77], fuzzy goal programming techniques [28,58], and Pareto concept [4,8,16(Ch. 4),17,26,31-34,57].

Various conventional methods have been presented to solve the RPP problem and assured their incompetence in handling multi-objective nonlinear problems and they may converge to a local optimum. MOAs that mimic the nature opened a new era in computation. For the past decades, numerous research applications of MOAs have been concentrated for solving the RPP problem. In this particular area, the research is still young which broadens the scope and viability of MOAs exploring new modifications and developments in solving the RPP problem. This paper presents a broad overview of solution methods for solving the RPP problem which are analytical approaches, arithmetic programming approaches, and meta-heuristic optimization techniques. Also, the different applications of meta-heuristic algorithms for solving the RPP problem are extensively reviewed and thoroughly discussed. Furthermore, the multi-objective formulations of reactive power planning and operations are studied to clarify their merits and demerits. This paper is organized as follows. The formulation of the RPP problem is presented in Section 2. Section 3 discusses the different methods applied to solve the RPP problem. The multi-objective formulations of the RPP problem are discussed in Section 4. The concluding remarks are highlighted in Section 5.

2. General formulation of the RPP problem

The purpose of the RPP problem is to determine "where" and ''how many" new VAR compensators must be added to a network for a predefined horizon of planning at minimum cost while satisfying an adequate voltage profile during normal conditions and contingencies. Fig. 1 illustrates the flowchart of the RPP problem.

After defining the system data, the generation/load patterns are developed for a predefined horizon of planning. Then, the optimal locations of new reactive power sources are identified. They may be selected according to certain indices or all load buses may be considered as candidate buses [14,15].

After that, the control variables (RPP variables) are optimized to achieve certain objective functions subject to set of equality and inequality constraints. Control variables include generator bus terminal voltages, reactive power generation of existing and new VAR sources and transformer tap ratio.

Define Bus, Branch, and generation System data

Develop Generation/Load

Identify the optimal locations of additional VAR sources

Optimize the RPP variables considering certain objectives

Figure 1 Flowchart of the RPP problem.

The generator bus voltages are continuous in nature, while both reactive power generation of existing and new VAR sources and transformer tap ratio are discrete. The dependent variables include load bus voltage magnitude, active power generation at slack bus, the power flows through the transmission lines, and reactive power outputs of the generators.

There are various objective functions that have been utilized in the RPP problem such as minimization of VAR investment cost and system operational cost of real power losses, improvement of voltage profile, and enhancement of voltage stability. However, the modeling of each objective has different shapes. Conventionally, the classical objective of the RPP problem is to achieve the minimum investment cost of additional reactive power supplies and minimize the system operational cost of power losses [1,11,24,25,28,35-38,50,51,53,56,68] as follows:

Min F = Min (Ie + Oc) (1)

where IC is the investment cost of new reactive power supplies and OC is the operational cost of power losses. The investment costs of VAR sources can be generally modeled with two components, a fixed installation cost at bus i (e,) and a variable purchase cost of capacitive or inductive source at bus i (Cci| Qci|), [16,24-26,28,31,34,35,37,38,50,51,53,56,68] as follows:

Ic = £> + CciQ1) (2)

where Nc is the reactive compensator buses. This model requires considering the reactive power devices to be already installed before the optimization for its size. On the other hand, another general model of IC has been used as [1-3,27,43]:

Ic = £ (e + Cci \Qa 0 be (3)

where Nb is the total number of busses, and bC is the binary decision variables for installing capacitive source. Although the complexity of using binary variables to indicate whether the VAR source will be installed, this model will give a chance

to consider all load buses to be candidates to install new reactive power sources. Traditionally, the annual cost of energy losses has been used as a direct measure to the operational costs (OC) [1,16,24,25,28,31,34-38,51,53,56,68] as follows:

Oc = ¡^dLP\

where h is the per unit energy cost, dL is the duration of load level (h), Nl is the number of load level duration, and PLoss s the real power loss during the period of load level L. On the other side, the minimization of network transmission power losses (Ploss) has been sometimes used directly instead of converting it to operational costs in the reactive power operation [4,29,32,39-41,46,47] and planning [20,43,44,52]. Also, the power system has to satisfy equality and inequality constraints corresponding to the load flow model and operational variables as follows:

Qgi - Qu + QC + Qct - Vj(Gi sin hj - Bj cos hj) = 0,

i = 1,2, ...Nb

- PU - V^Vj(Gj cos hj + Bj sin hj) = 0,

j=1 ■Nb

ОШГ 6 QgI 6 QT7, i = 1,2,......Npv

г"1" 6 Vi 6 V™, i = 1,2,......Nb

ТШn 6 Tk 6 ТШ", k = 1,2,......Nt

6 S™, L = 1,2,......Nl

0 6 Qce 6 Qmax, e = 1,2,......Nc

0 6 QnCj 6 QT7W ■ ßc j, j e candidate buses

Pmn Ps

6 Ps 6 Pm

(8) (9)

(10) (11) (12) (13)

where V and Vj are voltages at buses i and j, respectively; 6y is phase angle between buses i and j; Gj and By are mutual conductance and susceptance between buses i and j, respectively; (Pgi — PLi) and (Qgi — QLi) are the net real power injection at bus i, and the net reactive power injection at bus i, respectively; QCi is the capacitive or inductive power of existing VAR source installed at bus i. QCi refers to the capacitive or inductive power of new VAR source installed at bus i. Qgi is the reactive power output of a generator i, and Npv refers to the total number of voltage-controlled buses. Vi is the voltage magnitude of bus i. Tk is the tapping change of a transformer k, and Nt refers to the total number of on-load tap changing transformers. Sflow refers to the apparent power flow, Smax is the maximum MVA rating of the transmission lines and transformers, and NL refers to all transmission lines in the system. QCe is the reactive power output of existing VAR source at bus e, Qmax s its maximum capacity, and NC refers to the total number of existing VAR sources. n refers to the new installed VAR sources, and pC is always equal 1 for the investment cost

of VAR sources modeled in Eq. (3). Ps is the active power generation at the slack bus.

3. Solution methods for the RPP problem

RPP is a nonlinear multi-objective constrained combinatorial optimization problem for large power systems with a lot of uncertainties. Generally, the RPP problem has been solved by analytical approaches, arithmetic programming approaches, and meta-heuristic optimization techniques. Fig. 2 depicts the family and subcategories of the solution algorithms for the RPP problem. As shown, the several applications of meta-heuristic algorithms are subcategorized into evolution based, and swarm intelligence [78]. Added to that, hybridization between different algorithms is taken into consideration to improve the solution quality.

3.1. Analytical approaches

Analytical approaches are very important to understand the different effects and benefits of the location and size of reactive power sources [11-13]. The issues of RPP have been analyzed with reactive power pricing in [11] where a trade-off between the transmission loss and installation cost of new capacitors has been executed incorporating detailed hourly loading conditions. In [12], three economic benefits with assumption of a constant VAR injection and a fixed location have been analyzed. These benefits include reducing losses, shifting reactive power flow to real power flow, and increasing the transfer capability. The economic benefits have been updated by executing a set of optimal power flow (OPF) runs. Also, the reactive market-based of economic dispatch has been addressed in [13]. However, the benefits to the utilities from the allocation, installation, and operation of VAR compensators have not been discussed. Analytical approaches lend a lot of information and clear vision about the economic and technical benefits under different scenarios. They are quite helpful to design future framework of reactive power management and pricing for different players in the deregulated environment. On the other hand, they are time-consuming and may not be suitable for medium and large-scale power systems. Analytical approaches are as accurate as the model developed. They are based on its corresponding OPF which has been usually solved using nonlinear algorithms such as Modular Incore Nonlinear Optimization System (MINOS) [11-13] using General Algebraic Modeling Systems (GAMS) procedures [79].

3.2. Arithmetic programming approaches

Arithmetic programming approaches are also called Conventional Optimization Algorithms (COAs). A variety of conventional methods have been widely used to solve the reactive power operation and planning for years [14-16(Ch. 2),17(Ch. 3)]. COAs have been developed and implemented to solve the RPP problem. Table 1 shows a comparison between various COAs that have been applied to the RPP problem.

3.3. Meta-heuristic optimization algorithms

Meta-heuristic Optimization Algorithms (MOAs) are extensively used in solving multi-objective optimization problems

i = 1,2,

Figure 2 Family and subcategories of the solution algorithms for the RPP problem.

since they can find multiple optimal solutions in a single run. Different MOAs are applied efficiently to solve the RPP problem. Table 2 shows a comparison between various MOAs that have been applied to the RPP problem.

Since, the settings of their key parameters have a large impact on their performance, the adaptive MOAs have been developed recently and applied to the RPP problem. Some of the adaptive MOAs reported are as follows: the IHS algorithm [44], Chaotic DE algorithm [17], JADE-vPS algorithm [6], adaptive model of IA [8], EPSO [10,71], improved model of DE algorithm [42,80], SARGA [30], FAPSO algorithm [67], and MNSGA-II [31-33,76]. Although the adaptive models of MOAs reduce the complexity of parameter selection, the selected adaptation strategy influences on their performance and they have a high computational burden that needs more calculations to adapt the parameters.

4. Multi-objectives treatment of the RPP problem

In recent years, the RPP problem has been formulated as multi-objective optimization problem. Several methods have been presented to handle the multi-objective formulation of the reactive power planning and operation problems.

4.1. The mathematical sum approach

Multi-objective RPP problem has been treated using the mathematical sum approach as in Eq. (1) to minimize both the investment and operational costs [1,11,24,25,28,35-38,51,53,5 6,68]. Although this model is very simple, it doesn't prefer any objective over the others. Also, it is restricted where the multi-objectives should be with the same nature as in Eq. (1); both objectives are in the same kind (costs in dollar), else it will be meaningless.

4.2. The weighted sum approach

Multi-objective RPP problem has been treated also using weighted objective functions [27,29,40,43,44,47,69]. Weighted sum of different objectives can be generally modeled as follows:

Min F — Min xiFi where ^x — 1 (14)

j=1 i—1

where x and F are the weighting factor and the objective function for each goal i, respectively and NF is the total

Table 1 Comparison among COAs implemented to solve the RPP problem.

Category

Remarks

Merits

Demerits

NonLinear Programming [11-13] (NLP) method Modular Incore Nonlinear Optimization System (MINOS) solver*

Mixed Integer NonLinear [19] Programming (MINLP) solver (KNITRO 8 solver)

Interior Point (IP) method [3]

Discrete and Continuous OPTimizer (DICOPT)

solver

Penalty Successive Conic Programming (PSCP) method

Dual Projected Pseudo Quasi-Newton (DPPQN) method

Branch and Bound (B&B) method

[4,18]

[21,23]

> MINOS employs a project Lagrangian algorithm with a linear approximation to the nonlinear constraints. It then uses the reduced-gradient algorithm for solving a linearly constrained sub-problem with a sequence of iterations

> The load uncertainty and different contingencies have been considered in multi-scenarios extracted using a scenario tree reduction methodology. KNITRO implements the interior method where, the nonlinear programming problem is replaced by a series of barrier sub-problems

> RPP problem has been formulated as a stage of TEP problem

> The candidate buses to install VAR sources have been selected based on L-index as a voltage stability index

> DICOPT solves a series of NLP and Mixed Integer Programming (MIP) sub-problems. It is based on outer approximation of the objective function, equality relaxation, and augmented penalty of the inequality constraints and the objective function. PSCP method is generally a linear program with an additional nonlinear conic constraints corresponding to multiple state constraints as a penalty function

The PSCP algorithm has been solved by polynomial time primal-dual IP methods to find a common value of the decision variables in each state in a successive manner

This method considered only power losses as a single objective RPP problem

The investment cost for reactive power sources has been handled as budget constraint

This method employed a sequence of MIP method where, sensitivities of voltage stability margin and voltage magnitude have been used in this RPP formulation

> In B&B, the search continues by creating two new sub-problems, each one is then solved by the same procedure, resulting in a search-tree of subproblems

> Fast computation performance. It solves quickly a large number of single optimizations which corresponds to different loading and contingency conditions

Fast computational performance.

> Very suitable to handle with both continuous and discrete variables

> No need for calculating 1st or 2nd derivatives of the nonlinear objectives or constraints

Iterative approach for computing steps

Suitable for solving the RPP problem as a MINLP problem that involves integer variables and continuous variables

Fast computation performance Very fast computational method This method handled with outage scenarios and different load levels under voltage profile and stability constraints

Fast computational technique Efficient for solving RPP problems

No need for restarting the tree search and only a single tree is required It is fast

It provides good solutions for large-scale power systems

It is based on simplifications of sequential linearization

It is highly dependent on choosing the starting point

It finds locally optimal solutions It could be trapped in a local optimum and there is no guarantee to find the global optimum even if you run the algorithms for infinite long because the diversity of the solutions is limited The multi-objective functions have been treated mathematically sum for each scenario in [19] Neglecting the effect of transformer tap changing on the RPP problem in [3]

It does not necessarily obtain the global optimum

It is based on linear approximations of nonlinear functions at each iterations and accumulating them due to outer-approximations The solution of each conic program employs a linearization of the power flow equations at the current operating point High computational burden due to multiple states VAR planning includes outage scenarios and different load levels

It becomes too slow if number of variables is large

It ignored the effect of generator voltages and tap changing transformers considering only the VAR patterns as control variables

> Complex and high computational burden due to many levels and load cases

> The formulation has been approximated to be linear using voltage stability margin sensitivities and voltage magnitude sensitivities

> It finds locally optimal solutions

MINOS solver [11-13], and DICOPT solver [4,18] have been formulated in GAMS software [79].

Table 2 Comparison among MOAs implemented to solve the RPP problem.

Category

Remarks

Genetic [27-

Algorithm (GA) 29.50]

[24.25]

[31-33.66 (Ch. 8)]

> The chromosomes are coded as binary bit strings This model is called Simple GA (SGA)

1 In [29]. SGA has been applied to solve the reactive power dispatch based on the Fuzzy Goal Programming (FGP) to minimize the weighted sum of membership goals

> The RPP problem has been formulated in a stochastic model which represented the uncertainties of generator outputs and load demands with specified probability distributions

> SGA based on Monte Carlo simulation has been used as a solution tool to minimize both the costs of energy loss and investments of new VAR sources

> The chromosomes have been coded as a finite-length string of real numbers This model is called real coded GA (RGA)

> Blend crossover (BLX-a) and normally distributed mutation operators have been applied directly to real values

> A self-adaptive model of real coded genetic algorithm (SARGA) has been presented to solve the optimal reactive power dispatch (ORPD) problem The simulated binary crossover (SBX) operator has been used to create offsprings relative to the difference in parent solutions

> Representation of both binary and real variables has been deemed This improved GA carried out the uniform mutation operator to the mixed variables with some modifications, the blend crossover operator (BLX-a), and simple crossover for real and integer parts, respectively

> Non-Dominated Sorting Genetic Algorithm II (NSGA-II) has been utilized to solve the multi-objective RPP to minimize the investment costs of shunt compensation and the average load bus voltage deviation as well NSGA-II ranks the individuals based on the concept of Pareto non-dominance

■ A Modified NSGA-II (MNSGA-II) has been applied to the RPP problem In [31.33]. Pareto-front has been created by converting the multi-objectives into single one using conventional weighted sum method and varying the weighting

Capabilities

• It involves a high degree of randomness

• Good diversity of the solutions to avoid being trapped in a local optimum

• Easy to use

• The violation probability shouldn't exceed a chosen confidence level

• Different planning schemes have been presented by altering the confidence levels of the objective and constraints

Demerits

• Slow convergence rate

• No guarantee that GA will find a global optimum

• Some difficulties in chromosome encoding

• It is highly dependent on crossover and mutation rates

• The voltage constraints may be violated in some exceptional cases

• The most appropriate choice hasn't been determined

• The effect of tapping change of transformers hasn't been considered in the model

> It can find the global optima as the number of iterations approaches infinite

> Easy to be modified and joined with other approaches

> In this type of crossover, close-parent solutions are monotonically more likely chosen as offspring than solutions distant from parents

Since BLX-a is based on the interval process for real variables, the new off-springs depend on the location of both parents and so they will be close to the parents if both parents are close to each other, and vice versa [5]

It is highly dependent on crossover and mutation rates and effect on stability and convergence It finds sub-optimal solutions

Design for binary and real search spaces

Updating Pareto set using a Crowding Distance (CD) operator

More diversity of non-dominated solutions

Lateral diversity is lost More computational complexity

Dynamic modification of Pareto set using Dynamic Crowding Distance (DCD) High uniformity and maintains good diversity since the lowest DCD individual has been removed every time and DCD has been recalculated for the remaining individuals

High computational complexity In [33], the best compromise solution hasn't been included and the obtained Pareto front has been considered to give more choices to the decision

Table 2 (icontinued)

Category Ref. Remarks

factors randomly. In [31.66 (Ch. 8)], the best compromise solution among Pareto-optimal solutions has been determined based on TOPSIS method

• In [32]. MNSGA-II has been employed to the ORPD to minimize the real power losses and maximize the voltage stability using the L-index. In this paper, multiple runs of single objective optimization with weighted sum of objectives have been used to obtain Pareto-set

Differential [34- • DE algorithm has been used to solve the RPP

Evolution (DE) 37.25.38] problem to minimize both the VAR and energy

loss costs

• In [34.35]. the discrete variables have been treated as continuous and then rounding it to the nearest integer

• In [36]. the RPP problem has been formulated as a contingency constrained optimal RPP problem. The single line contingency analysis firstly has been used to identify the severe state and its voltage violated buses. Then, these voltage violations have been added as an additional constraint to the base RPP problem

• In [37.38]. Fast Voltage Stability Index (FVSI) has been used to identify the weak buses for the RPP problem which has been solved using DE algorithm

[39.40] • A multi-objective reactive power and voltage control problem has been solved by DE approach. In [39]. the candidate buses for VAR injection have been selected based on L-index to minimize real losses, voltage deviation and voltage stability index (L-index)

• In [40]. the power losses and the voltage deviation have been minimized

[41] • DE algorithm has been implemented to achieve

losses minimization, voltage profile improvement, and voltage stability enhancement

[52] • DE algorithm has been tested to solve the RPP

problem, including the placement and sizing of TCSC devices. The main factor to determine the optimal location of the TCSCs has been the loss reduction while, voltage stability enhancement, and voltage deviation reduction have been added as penalty terms

maker. Otherwise, the effect of the existing reactive power sources has been ignored in this reactive power dispatch model

• It can find near optimal solution regardless the initial parameter values

• Efficient method where it cannot be easily trapped in local minima

• Suitable convergence speed

• Robust

• It uses few number of control parameters

• Simple in coding and easy to use

• Easily handling integer and discrete optimization

• Very suitable to solve multi-dimensional function optimization as the RPP problem

Efficiency is very sensitive to the setting the control parameters. It is dependent on three main parameters which are population size (Np), mutation rate (F), and crossover rate (CR) Parameter tuning mostly by trial-and-error Crossover has the potential to destroy the directional information provided by the difference vectors for the sake of increasing diversity The convergence is unstable with a small population size

It may drop in local best

• Severe line outages have been taken into consideration to improve voltage stability

• Handling the RPP problem as a single objective optimization problem

• It considered real power of generators as decision variables which have more effects on losses

• More complex by solving both P and Q optimization problems in a single step

Table 2 (icontinued)

Category Ref. Remarks

Immune Algorithm (IA)

Seeker

Optimization Algorithm (SOA)

[46.48]

Harmony search [43] algorithm

■ A Self Adaptive DE (Chaotic DE) algorithm has been implemented to the RPP problem. It changes the mutation and crossover parameters to be updated each generation

> An improved model of DE has been presented to minimize both the energy loss and the installation costs while, the critical lines and buses to install FACTS controllers have been determined based on FVSI

> A new adaptive DE algorithm called (JADE-vPS) has been applied to minimize the total fuel cost with satisfying a minimum voltage stability margin for the optimal power flow. In this paper, an adaptive penalty function has been introduced where the penalty coefficients has been altered automatically from data gathered from the search process

> IA has been implemented in adaptive model to solve the reactive power flow in order to minimize power losses, voltage deviation, and enhance static voltage stability. Crossover rates, mutation rates and clone rates have been used all adaptive to change automatically at each generation related to the global affinity function

■ In [46]. SOA has been executed to the ORPD problem to minimize the real losses as a single objective function. In [48]. SOA has been implemented to minimize the power losses, voltage deviation and increasing voltage stability using L-index. This ORPD has been handled as minimizing different single objective functions

> A multi-objective reactive power control has been addressed using SOA. In this paper, the multi-objective functions were to minimize the transmission loss and voltage deviations while the voltage stability margin would be maximized by minimizing the eigenvalue of the non-singular power flow Jacobian matrix

> HS method has been used to determine the locations and the outputs of Static VAR Compensators (SVCs) to minimize the total investment costs, average voltage deviation and total system loss

• Self adaptation of mutation and crossover rates to improve efficiency

• The mutation factor has been changed dynamically instead of being constant as in the classic DE model

• Efficiency is still sensitive to population size

• More computational burden

• Not only mutation factor and crossover rate have been already self-adapted, but also population size has been automatically adapted in a very similar manner to the other two parameters.

• High computational burden and complexity

• Adaptive parameters avoid premature conver- • More computational burden and complexity gence and falling into a local optimal solution

• Good efficiency and convergence

• Easy to understand

• Suitable performance in balancing global search ability and convergence speed

• Although SOA handled only continuous variables. Refs. [46.47] tackled this problem by searching in a continuous space, and then curtailing the corresponding dimensions of the seekers' real-values into the integers

• SOA may be stuck at a local optimum for multimodal functions

• SOA is heavily dependent on its structures and parameters

• The different objectives have been normalized to be treated as a single objective with weighting factors

• Such complexities to determine the weighting factors

« Simple in concept « Easy to be implemented « Suitable convergence speed

> It is dependent on three parameters which are harmony memory considering rate, pitch adjustment rate, and bandwidth vector

Table 2 (icontinued)

Category Ref. Remarks

[45] • HS algorithm has been to minimize transmission

loss, L-voltage stability index, and voltage deviation

[44] • An Improved Harmony Search (IHS) algorithm

has been carried out to reduce losses, installation cost, and achieve better voltage improvement by assigning the SVC placement and sizing Evolutionary [50,51] • EP and ES work on the basis of organic evolution programming models. In [50], the RPP problem has been

(EP) and evo- decomposed into P and Q optimization modules

lutionary and each one is solved iteratively using EP and

strategies evolutionary strategy

(ES) • In [51]. the RPP problem has been solved using

EP method considering the highest load buses to place the new VAR sources [54] • EP technique has been applied to solve two sepa-

rate RPP procedures which addressed the optimal reactive power dispatch and the optimal transformer tap changer setting [66(Ch. • A Covariance Matrix Adapted Evolution Strategy 6),67] (CMAES) has been employed to solve the RPP

problem. In [67], the RPP problem handled the voltage stability index (L-index) as an additional constraint with specified threshold • In [66 (Ch. 6)]. CMAES has been applied to solve RPP problem in hybrid (pool and bilateral coordinated) electricity market. In this chapter, different objectives have been considered which were the total production cost of real and reactive power and the allocation cost of additional reactive power sources (SVC) Ant Colony [55] • ACO algorithm has been hybrid with immune

Optimization algorithm to solve the problem of reactive power

(ACO) algorithm optimization to reduce only the transmission loss

Particle Swarm Optimization (PSO) algorithm

> The ORPD has been solved using ACO method to minimize the losses as a single objective function. Sensitivity parameters have been used to express objectives and dependent variables in terms of control variables and based on a modified model of fast decoupled load flow

• Dynamically Altering PAR and bw to eliminate the drawbacks of keeping constants in the HS model

• Simple and direct method to represent system variables

• More randomness

• Good diversity

• ES converges faster compared to EP

• EP is less likely to fall into a local minimum

• Each objective function has been handled separately as a single objective optimization

• More computational burden and complexity

• ES has a higher probability to fall into a local minimum

• No guarantee for finding optimal solutions in a finite amount of time

• Parameter tuning is needed

• Such a complexity in the system of mutations

• A single objective optimization has been implemented for minimizing only transmission losses

• Self-adaptation of the covariance matrix (CM) and the global step size during each generation to increase efficiency

• Due to its consistency, CMAES has been usually used to generate reference Pareto-front to compare the performance of other MOAs [31-33,66 (Ch. 8)]

• Slower convergence performance

• The adaptation process in CMAES is very complex and the computational burden of sophisticat-edly strategy parameters is very high

> Stochastic kind

> Inherent parallelism

> Adaptation capability

1 Using positive feedback

> Convergence is guaranteed

Using trial and errors to parameters initializations Its mathematical execution and analysis is difficult

slower convergence speed Linear approximation using sensitivities Each objective function has been handled separately as a single objective optimization problem

• Simple in concept

• Easy to be implemented

• Suitable convergence speed

• Slow convergence rate

• Trapping into local optima

Table 2 (icontinued)

Category Ref. Remarks

• PSO algorithm has been applied to find the optimal placement of FACTS devices based on the contingency severity index (CSI) values which consider single and multiple contingency

[2] • PSO method has been applied to the RPP prob-

lem as a second stage to minimize of the VAR investment costs. This model considered load uncertainties and the uncertainties of wind turbine output obtained by a probability distribution function (PDF) using MCS while the reliability has been taken into consideration

[56] • PSO algorithm has been used for solving the RPP

problem to minimize the operation cost and investment cost of reactive power sources

[57] • The RPP problem has been solved using PSO

technique incorporated with Pareto dominance to minimize real power losses and installation costs

[16 (Ch. . A Vector Evaluated PSO (VEPSO) method has been implemented on the multi-objective RPP problem

Artificial Bee [59] Colony (ABC) algorithm

Gravitational [60.61]

Search

Algorithm

Firefly [62]

Algorithm (FA)

A modified PSO method has been applied for scheduling of reactive power control variables to maximize the reactive power reserves. In this paper, the ^-constraint approach has been used to assure desired static voltage stability margin based on a proximity indicator ABC was inspired by the foraging behavior of honey bee swarm. It has been executed for handling the ORPD problem in deregulated power systems after assuming an already established real power market

GSA was based on Newton's law of gravity and motion. In [60]. it has been applied to the RPP using FACTS to minimize the losses and bus voltage deviations. In [61]. opposition-based GSA for population initialization has been presented to solve the ORPD problem

FA was based on swarm behavior and has many similarities with PSO algorithm. It has been applied to minimize the real power loss or the voltage deviations

Teaching [63]

• Efficient

• Having few parameters to be adjusted

• Less dependent on initial points

• It is dependent on the inertia weight and learning constants.

• Using trial and errors to parameters initializations

• The effect of tap settings hasn't been considered for more simplicity in handling the RPP problem as a stage of the TEP problem

Handling the integer variables has been done by rounding it to the nearest discrete after relating it as floating variable

A well-distributed Pareto front by adding an external archive to decide whether a solution can be stored or not. based on Pareto dominance

Good efficiency

A fuzzy based mechanism is employed to extract the best compromise solution over the trade-off front

Better efficiency where, a fly-back mechanism has been applied to enable any violated particle to fly back to its previous position

The state variables have been added to the objective as penalties, such complexity is existed to determine the penalty factors More computational burden and complexity to update the best positions based on the global best stored in the archive using crowding and roulette wheel selection

More computational burden and complexity to determine Pareto front that VEPSO generates two swarms where each one is based on an objective. and to extract the best compromise solution More computational burden to execute the flyback mechanism

It is as simple as PSO and DE with few control parameters such as colony size and maximum cycle number

It is robust against initialization

It has the ability to explore local solutions

It is simple and easy to implement

It has a high randomness of the individual moves.

Thus, it provides the global exploration in the

search space

FA is simple and easy to implement It is good at exploration

It includes the self-improving process with the current space and it improves its own space from the previous stages

• ABC has poor exploitation characteristics

• Its convergence speed is also an issue in some cases

• It may get stuck in local optimum

• The local search ability of GSA is weak

• In [60]. it isn't robust against initialization. This feature is improved in [61]

• In [60.61]. the considered problem was formulated as a single objective optimization problem

• FA often traps into local optima

• The minimization of power losses or voltage profile improvement is handled as a single goal optimization

• Its parameters were set fixed and they do not change with the time

TLA often converges to local optima

Table 2 (icontinued)

Category Ref. Remarks

Learning Algorithm (TLA)

Chemical Reaction Optimization (CRO)

Water Cycle

Algorithm

Differential Search Algorithm (DSA)

• TLA was based on the simulation of a classical learning process which composed of two phases: (i) learning through teacher and (ii) learning through interacting with the other learners. It has been applied to handle the ORPD problem considering only the power loss

• CRO was based on the various chemical reactions occur among the molecules. It has been applied to the RPP using FACTS to minimize the transmission loss, improve the voltage profile and voltage stability

[65] • WCA is inspired from nature and based on the

observation of water cycle and how rivers and streams flow downhill toward the sea in the real world. It has been applied to minimize the weighted sum of the losses and the voltage deviations

• DSA was inspired by migration of super-organ-isms utilizing the concept of Brownian like motion. It has been applied to solve the non-feasi-bility problem solution of the fuel cost minimization problem (for a given operating point) by optimizing the RPP problem

• The candidate placements of VAR sources have been selected based on FVSI

Hybrid [16 (Ch. • A hybrid PSO-DE algorithm has been impie-

techniques 5),68] mented for solving the reactive power control

problem in electricity market

• PSO-DE algorithm carried out a differential operator from DE in the update of particle velocity of PSO

• A hybrid PSO-GA algorithm has been implemented to minimize the cost of reactive power generation, reactive power compensators and active power losses. BLX-a, and uniform mutation operators from GA algorithm are applied on the PSO particles

[69] • Another model of hybrid PSO-GA has been per-

formed to search for the optimal placement of SVC. PSO algorithm is implemented firstly until

It has balanced global search ability and convergence rate.

It has a good capability for global and local searching

The exploration features need more support Power loss was the only considered objective

• CRO is easy to implement

• However. CRO behaves like a random search to traverse the whole solution space, which could confine the algorithm's search ability

• It is robust against initial seeds

• It is simple and easy to use

• It has few control parameters

• It has a good exploration features

• The local search needs more modifications since it may stick in local optima

• However, it has good robustness indices for solving the considered RPP in [64]. it is highly sensitive to the initial kinetic energy and the concerned loss rate

• Its local search ability of is weak

• It is often traps into local optima

• Its robustness and consistence need more uphold

• It has a good exploration feature in the search space to locate the region of global optimum

• Therefore, its convergence rate is fast but it is also a problem in some cases

• A selection strategy has been added that a particle is moved to a new location only if the new location yields a better fitness value

• Crossover and mutation are done if there is no change in the global position for a number of iterations to avoid premature convergence

The minimization of fuel cost or load voltage deviations is handled as mono-objective optimization in two separate levels

Transformer tap settings and VAR sources are treated as continuous variables Its exploitation of the optimal solution requires more support

DSA is still novel and further researches are necessary to be developed and improved Slow convergence rate

More computational burden and complexity Both algorithms are very sensitive to the setting of the control parameters

Using trial and errors to parameters initializations

• Simple hybrid model and easy to implement

• Good diversity

• Slower convergence performance

• More control parameters which needed to be tuned

• Using trial and errors to parameters initializations

Table 2 (continued)

Category

Remarks

Hybrid techniques

[70] [10.71]

its stopping iteration number is reached. Then. GA updates the population considering the last PSO population as its initial population

> IA has been combined with ACO algorithm composing a hybrid Artificial Immune Ant Colony Algorithm (AIACA) to minimize objective only the real power losses

> A Hybrid Evolutionary Programming method (HEP) has been executed to solve the RPP which combines EP technique as a base stage search toward the optimal region, and a direct search technique to reduce the size of search region to locally search for the global optimum. The fittest individuals in the combined population haven't been chosen in the next generation but they have greater chances than others

> A hybrid method combines the direct search, and PSO technique has been implemented to solve the ORPD. and compared with HEP method

> Evolutionary Particle Swarm Algorithm (EPSO) method has been applied to the reactive power control and planning. EPSO formulation is based on the particle movement like the classical PSO where, the weights are mutated using EP mutation factor

> A hybrid between fuzzy reasoning approach and PSO method has been introduced. Fuzzy membership of loss sensitivity at each bus has been evaluated to determine candidate buses to install shunt capacitors. PSO has been used immediately to minimize the investment costs and transmission losses as well

■ A Fuzzy Adaptive PSO (FAPSO) method has been presented for solving the problem of reactive power and voltage control. A fuzzy optimization approach based on pseudo-goal function has been used to convert the different objectives, which were the active power loss, voltage deviation and the voltage stability index, into a single-objective optimization problem. Then, this single-objective optimization problem has been solved using the FAPSO approach

[72.74]

• It makes use of the positive feedback principle of ACO method and the rapidity of IA to avoid trapping into a local optimal solution

• The direct search technique tackles difficulties in a fine-tuning of local search in EP method by direct searching toward the optimal region

• Reducing the size of search region

• Finer convergence and improving the solution quality

• Handling the reduction of the transmission loss as a single objective optimization problem

• Slow convergence rate

• The direct search technique is very dependent on the initial starting point

• Slower convergence speed

• Parameter tuning is needed

• Such a complexity in the system of mutations

• Handling a single objective optimization problem which is the real losses

• More diversity of solutions • Such a complexity due to EP mutations

• Considering different contingencies and load • Parameter tuning is needed levels in [71]

• Simple model as it provides two different levels where, fuzzy memberships are used for capacitor placements and the control variables are handled by PSO technique

• It is still dependent on the inertia weight and learning constants

• It may trapped into local optima

• Using trial and errors to parameters initializations

• In FAPSO approach, the inertia weight and the learning coefficients have been dynamically varied by fuzzy rules based on the fitness values of particles during optimization process

• More complexity of representing fuzzy memberships

• More computational burden to adapt PSO parameters

• Slow convergence rate

Fast computational performance Handling easily conflicting objectives

■a a H

S & ^ S

2 û a °

& ^ sd OJ >

§ 3 6

^ s s ^ £

-B O ^

" u Ê

S Ji •S —

^ H, a

B eg ja

s n.U <5 g >

O S—i

£ „

1—' ft

ETL <2

G ÇU £

¿S «3 «

C <u 2

-B tS <3

< C3 -O

S? s1'

S— qj

S C3 O -B ft

■3 b

oj > o

> B '3 TS

=a 3 's

^ .S B

_B ~ °

■M M

'g TS -B

^ ft 3

S -S -s' s ° S g s

S o M o

5 H 40 ÎJ

-B g a <5 <

c« Ph

ft S P Q ^

n u 2 Ph M -B

o ^ <U

^ -B B <

J> TS Ô «

OJ ^ m

number of objective functions. A normalization process has been incorporated to the weighted sum approach in [44,47]. In [44], each objective function (real power losses, voltage deviation, and VAR investment cost) has been normalized in a comparative manner with its base case value. Also, the normalization process can be done as a fuzzification process [47] to map all objectives within the range of [0,1]. Then it is generally modeled as weighted sum defined in Eq. (14). The normalization process enables comparing the different objectives in a fairly manner. The optimal solution is greatly affected by the selection of the weights. Another problem associated with this approach is that it may find solutions that are close to one or more operating constraint violations [26].

4.3. The e-constraint approach

The e-constraint approach has been used in tackling multi-objective problems of reactive power planning and control [6,18,20,43,76,77]. This method optimizes the main objective (Fm) as a single objective optimization problem while, it considers other objectives as constraints restricted by some chosen threshold levels.

Min Fm while Fi 6 ei

' (15)

i — 1,2,3,......Nf, i—m

where ei is a threshold level specified by the user for each objective (Fi). Choosing et is easier than choosing adequate values for weight factors (ra), but the optimal solution still depends on its value. In [20], the capacitors has been installed to minimize the real losses (main objective) while its investment cost has been handled with budget limit (e-constraint). Also, the loading parameter (k) has been a constrained to guarantee a minimum voltage stability margin in [6,18]. In [9], Schur's inequality has been used to assure required static voltage stability margin. The eigenvalue analysis has been used as a stability margin proximity indicator where a threshold value of proximity indicator must be specified for secure operation. Also, the objective of enhancing the voltage stability has been achieved by restricting the static voltage stability index (L-index) by a maximum level [76,77].

4.4. The fuzzy goal programming approach

Fuzzy Goal Programming (FGP) has been presented in [29,67] for solving the problem of reactive power and voltage control. The active power loss, voltage deviation and the voltage stability index (L-index) have been converted into a single-objective optimization problem. In [29], GA has been employed as a solution tool to the FGP formulation to minimize the weighted sum of membership goals. Fuzzy adaptive particle swarm optimization (FAPSO) approach has been implemented based on the maximumminimum value of all membership functions of the objectives and constraints [67]. The main advantage of the FGP formulation is treating the multi-objective as a single objective optimization problem effectively without selecting weights or thresholds as in the weighted sum or e-constraint methods, respectively.

4.5. The Pareto optimality approach

Multi-objective RPP problem has been achieved using the concept of Pareto-optimality [4,8,16(Ch. 4),17,26,31-34,57]. The solution is said to be Pareto-optimal if there is no a better solution in terms of all objectives.

4.5.1. Methods of creating Pareto front

Meta-heuristic algorithms typically generate sets of solutions, allowing computation of the Pareto set based on the non-dominance concept [8,26,57]. Also, Pareto-front has been created using various runs of single objective optimization with varied weight factors of different objectives [31-34,76(Chs. 7 and 8)]. The e-constraint method has also been implemented with Pareto optimal front where the specified bounds of objective constraints are changed to get the Pareto front [4]. However, this method is time-consuming and tends to find weak non-dominated solutions in Pareto front since it depends on the objective bounds specified by the user. Moreover, Vector Evaluated PSO (VEPSO) method has been used to solve the multi-objective RPP problem to minimize the operational and installation costs and the voltage stability index (L-index). VEPSO determines Pareto front by generating two swarms, one swarm for each objective [16(Ch. 4)]. The strength of Par-eto Evolutionary Algorithm (SPEA) has been used for solving the multi-objective RPP problem to minimize the real power loss and the bus voltage deviations [17]. It firstly stores the non-dominated solutions in an external Pareto set to give scalar fitness values (strength) to individuals. Then, it uses clustering approach to reduce the Pareto set when the number of the non-dominated solutions exceeds the pre-specified value. The fitness (strength) of any individual is calculated based on only the solutions stored in the external Pareto set. The selection operator is applied to the population individuals and all solutions in the external Pareto set.

4.5.2. The best compromise solution over Pareto solutions Determination of a single optimal solution that simultaneously optimizes all multi-objective functions is difficult. However, the decision makers can perform a trade-off analysis and select among the set of the non-dominated solutions [33,34,57]. The fuzzy decision-making tool has been presented to determine the best compromise solution for the RPP problem [4,16(Ch. 4),17]. Each objective Ft is fuzzified with a membership function i as in Eq. (16) and Fig. 3 shows its related fuzzy modeling. Then, the best solution is selected, which achieves the maximum membership ik which is defined in Eq. (17) or the maximum normalizing membership ik which is defined in Eq. (18) [4]:

Fi = lÁFi) =

/max— Fi

Fmax_ /mm 0

Fii < F"111 Fmin 6 F¡ 6

F, > F"ax

ELE Mii( FO EM^ïk)

En s^M

Figure 3 Fuzzy membership model for objective functions.

where k refers to each non-dominated solution, M is the number of objectives, n is the total number of the non-dominated solutions, and mi refers to weight value of the ith objective function. This method suffers from the problem of how to select the weight values m¡. In [4], the weight values mi has been selected based on the importance of economic and technical aspects. Moreover, the best compromise solution could be obtained using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method [31,32,76(Chs. 7 and 8)] as a multiple criteria decision making approach. In this technique, the relative performance of each non-dominated solution with respect to each criterion is identified and the geometric distance between each solution and the ideal solution in each criterion is calculated. Finally, the best compromise solution can be determined according to the maximum relative closeness to the ideal solution. In [32], TOPSIS approach has been used to rank the obtained MNSGA-II solutions for the reactive power dispatch to minimize two objectives, real power losses and L-index. The best compromise solution has been determined by a single decision maker. In [31], TOPSIS approach has been also used to find the best compromise for the RPP problem to minimize the combined operating and VAR allocation cost improves the voltage profile and enhances the voltage stability. In spite of its simplicity, TOPSIS approach does not take the relationships of different criteria into consideration. On the other hand, Pareto concept has been incorporated to the immune algorithm in [8] to define the partial affinity of an antibody (solution) to each antigen (objective). Then, the best compromise solution was based on the global affinity (sum of partial affinities).

5. Conclusion

Meta-heuristic optimization algorithms are going to be a new revolution in computer science. They opened a new era in the next generation of computation and optimization. In this paper, the solution algorithms of one of the widely significant optimization problems in electric power systems which is the RPP problem are extensively reviewed and thoroughly discussed. They are categorized into analytical approaches, arithmetic programming approaches, and meta-heuristic optimization techniques.

Analytical approaches present detailed information about the installations of reactive power compensators and its economic and technical benefits under different scenarios. They are quite helpful to design future framework of reactive power management and pricing for different players in the deregulated environment. However, they are time-consuming and may not also be suitable for medium and large-scale power systems. They are as accurate as the corresponding OPF model.

Arithmetic programming approaches have been widely used to solve the reactive power operation and planning for years. They are usually based on some simplifications such as sequential linearization and using the first and second differentiations of objective function and constraints. They may converge to a local optimum. They are very weak in handling multi-objective nonlinear problems. On the other side, they have fast computation performance and thus they provide the capability to solve a large number of single optimizations associated with different loading and contingency conditions.

An overview of a range of MOAs drawn from an evolutionary based or swarm intelligence is presented including GA, DE, HS, SOA, EP, ACO, IA, PSO, ABC, GSA, FA, TLA, CRO, WCA, and DSA. Each algorithm is distinguished with different features. Generally speaking, they perform with heuristic population-based search strategies that involve stochastic variation and selection. They are very suitable in solving multi-objective RPP problem. They are robust, effective, consistent, and can find multiple optimal solutions in a single simulation run.

Particularly, the scope of this area is really vast and there are great opportunities in applying novel approaches/algorithms to solve the RPP problem. Moreover, hybridization of different techniques is another research area to make use of different advantages to improve the quality of solution of the RPP problem. Otherwise, the adaptive strategies of MOAs to the strategic parameters are required to reduce the complexities of its selection.

Also, the multi-objective reactive power planning and operation are discussed to clarify their merits and demerits. Meta-heuristic algorithms typically generate Pareto set based on the non-dominance concept. Also, Pareto-front can be created using the conventional weighted sum or the e-constraint method.

References

[1] Rahmani M, Rashidinejad M, Carreno EM, Romero RA. A combinatorial approach for transmission expansion & reactive power planning. In: Proceedings of IEEE/PES transmission and distribution conference and exposition, Latin America; 2010. p. 529-36.

[2] Hemmati R, Hooshmand R, Khodabakhshian A. Market based transmission expansion and reactive power planning with consideration of wind and load uncertainties. Renew Sustain Energy Rev 2014;29:1-10.

[3] Mahmoudabadi A, Rashidinejad M. An application of hybrid heuristic method to solve concurrent transmission network expansion and reactive power planning. Int J Electr Power Energy Syst 2013;45(1):71-7.

[4] Ara AL, Kazemib A, Gahramania S, Behshada M. Optimal reactive power flow using multi-objective mathematical programming. Sci Iran 2012;19(6):1829-36.

[5] Devaraj D, Roselyn JP. Genetic algorithm based reactive power dispatch for voltage stability improvement. Electr Power Energy Syst 2010;32(10):1151-6.

[6] Nakawiro W, Erlich I. A new adaptive differential evolution algorithm for voltage stability constrained optimal power flow. In: Proceedings of the 17th power systems computation conference, Sweden; 2011. p. 1-7.

[7] Abou El-Ela AA, Kinawy AM, El-Sehiemy RA, Mouwafi MT. Optimal reactive power dispatch using ant colony optimization algorithm. Electr Eng 2011;93(2):103-16.

[8] Hugang X, Haozhong C, Haiyu L. Optimal reactive power flow incorporating static voltage stability based on multi-objective adaptive immune algorithm. Energy Convers Manage 2008;49 (5):1175-81.

[9] Arya LD, Titare LS, Kothari DP. Improved particle swarm optimization applied to reactive power reserve maximization. Int J Electr Power Energy Syst 2010;32(5):368-74.

[10] V. Miranda, N. Fonseca. EPSO-evolutionary particle swarm optimization, a new algorithm with applications in power systems. In: Proceedings of IEEE/PES transmission and distribution conference and exhibition, Asia Pacific; 2002. p. 745-50.

[11] Chattopadhyay D, Bhattacharya K, Parikh J. Optimal reactive power planning and its spot pricing: an integrated approach. IEEE Trans Power Syst 1995;10(4):2014-20.

[12] Zhang W, Li F, Tolbert LM. Analysis of var benefits with application to var planning. In: Proceedings of the 8th international power engineering conference (IPEC), Singapore; 2007. p. 146-52.

[13] Li F, Zhang W, Tolbert LM, Kueck JD, Tom Rizy D. A framework to quantify the economic benefit from local var compensation. Int Rev Electr Eng 2008;3(6):989-98.

[14] Zhang W, Li F, Tolbert LM. Review of reactive power planning: objectives, constraints, and algorithms. IEEE Trans Power Syst 2007;22(4):2177-86.

[15] Zhang W, Tolbert LM. Survey of reactive power planning methods. In: Proceedings of IEEE power engineering society general meeting, San Francisco; 2005. p. 1430-40.

[16] Rajkumar P. Application of particle swarm optimization technique for reactive power optimization problem. Ph.D. thesis. Faculty of Electrical Engineering, Anna University, Chennai; August 2010.

[17] Sutradhar S. Multi-objective reactive power planning using a new improved modified differential evolution algorithm. Master thesis. Faculty of Engineering and Technology, Jadavpur University, Kolkata; 2012.

[18] Chattopadhyay D, Chakrabarti BB. Reactive power planning incorporating voltage stability. Int J Electr Power Energy Syst 2002;24(3):185-200.

[19] Lopez JC, Contreras J, Mufioz JI, Mantovani JR. A multi-stage stochastic non-linear model for reactive power planning under contingencies. IEEE Trans Power Syst 2013;28(2):1503-14.

[20] Lin C, Lin S, Horng S. Iterative simulation optimization approach for optimal volt-ampere reactive sources planning. Int J Electr Power Energy Syst 2012;43(1):984-91.

[21] Liu H, Jin L, McCalley JD, Kumar R, Ajjarapu V, Elia N. Planning reconfigurable reactive control for voltage stability limited power systems. IEEE Trans Power Syst 2009;24 (2):1029-38.

[22] Jabr RA, Martins N, Pal BC, Karaki S. Contingency constrained var planning using penalty successive conic programming. IEEE Trans Power Syst 2012;27(1):545-53.

[23] Liu H, Krishnan V, McCalley JD, Chowdhury A. Optimal planning of static and dynamic reactive power resources. IET Gener Transm Distrib 2014;8(12):1916-27.

[24] Kumar SKN, Renuga P. Reactive power planning using real GA comparison with evolutionary programming. ACEEE Int J Electr Power Eng 2010;1(1):21-5.

[25] Kumar SKN, Renuga P. Reactive power planning using differential evolution: comparison with real GA and evolutionary programming. Int J Recent Trends Eng 2009;2(5):130-4.

[26] Small SM, Jeyasurya B. Multi-objective reactive power planning: a Pareto optimization approach. In: Proceedings of international conference on intelligent systems applications to power systems; 2007. p. 1-6.

[27] Boroujeni BS, Boroujeni SMS, Memaripour A. Reactive power expansion planning under a deregulated market power system. Res J Appl Sci, Eng Technol 2012;4(19):3755-9.

[28] Yang N, Yu CW, Wen F, Chung CY. An investigation of reactive power planning based on chance constrained programming. Int J Electr Power Energy Syst 2007;29(9):650-6.

[29] Baran Pal B, Biswas P, Mukhopadhyay A. GA based FGP approach for optimal reactive power dispatch. In: Proceedings of first international conference on computational intelligence: modeling techniques and applications (CIMTA); 2013. p. 464-73.

[30] Subbaraj P, Rajnarayanan PN. Optimal reactive power dispatch using self-adaptive real coded genetic algorithm. Electr Power Syst Res 2009;79(2):374-81.

[31] Ramesh S, Kannan S, Baskar S. Application of modified NSGA-II Algorithm to multi-objective reactive power planning. Appl Soft Comput 2012;12:741-53.

[32] Jeyadevi S, Baskar S, Babulal CK, Iruthayarajan MW. Solving multiobjective optimal reactive power dispatch using modified NSGA-II. Int J Electr Power Energy Syst 2011;33 (2):219-28.

[33] Ramesh S, Kannan S, Baskar S. Application of modified NSGA-II algorithm to reactive power optimization. In: Proceedings of first international conference on computer science, engineering and information technology, India; 2011. p. 344-54.

[34] Cuello AA, Cedeüo JR. A differential evolution approach to optimal reactive power planning. In: Proceedings of IEEE/PES transmission and distribution conference and exposition, Latin America, Venezuela; 2006. p. 1-7.

[35] Cuello AA. Applications of the differential evolution optimization algorithm in power systems planning, operation and control. Master of science-electrical engineering (power systems), University of Puerto Rico, Mayaguez Campus; 2006.

[36] Rajkumar P, Devaraj D. Differential evolution approach for contingency constrained reactive power planning. J Electr Syst 2011;7(2):165-78.

[37] Vadivelu KR, Marutheswar GV. Soft computing technique based reactive power planning using NVSI. Int J Electr Syst 2015;11 (1):89-101.

[38] Vadivelu KR, Marutheswar GV. Fast voltage stability index based optimal reactive power planning using differential evolution. Electr Electron Eng: Int J 2014;3(1):51-60.

[39] Immanuel DG, Kumar GS, Rajan CCA. A multi objective differential evolution approach for optimal reactive power and voltage control. Indian Streams Res J 2013;3(8):1-8.

[40] Abdelmoumene M, Mohamed B, Boubakeur A. Optimal reactive power dispatch using differential evolution algorithm with voltage profile control. Int J Intell Syst Appl 2013;5(10):28-34.

[41] Abou El Ela AA, Abido MA, Spea SR. Differential evolution algorithm for optimal reactive power dispatch. Electr Power Syst Res 2011;81(2):458-64.

[42] Vadivelu KR, Marutheswar GV, kumar SKN. Artificial intelligence technique based reactive power planning incorporating facts controllers in real time power transmission system. ACEEE Int J Electr Power Eng 2014;5(1):31-7.

[43] Boroujeni SMS, Delafkar H, Behzadipour E, Boroujeni AS. Reactive power planning for loss minimization based on harmony search algorithm. Int J Nat Eng Sci 2011;5(2):73-7.

[44] Sirjani R, Mohamed A. Improved harmony search algorithm for optimal placement and sizing of static var compensators in power systems. In: Proceedings of first international conference on informatics and computational intelligence (ICI), Bandung; 2011. p. 295-300.

[45] Khazali AH, Kalantar M. Optimal reactive power dispatch based on harmony search algorithm. Int J Electr Power Energy Syst 2011;33(3):684-92.

[46] Dai C, Chen W, Zhu Y, Zhang X. Seeker optimization algorithm for optimal reactive power dispatch. IEEE Trans Power Syst 2009;24(3):1218-31.

[47] Dai C, Chen W, Zhu Y, Zhang X. Reactive power dispatch considering voltage stability with seeker optimization algorithm. Electr Power Syst Res 2009;79(10):1462-71.

[48] Shahbazi HR, Kalantar M. The optimal reactive power dispatch using seeker optimization algorithm based different objective functions. J Basic Appl Sci Res 2013;3(8):232-40.

[49] Lai LL. Application of evolutionary programming to reactive power planning - comparison with nonlinear programming approach. IEEE Trans Power Syst 1997;12(1):198-206.

[50] Lee KY, Yang FF. Optimal reactive power planning using evolutionary algorithms: a comparative study for evolutionary programming, evolutionary strategy, genetic algorithm, and linear programming. IEEE Trans Power Syst 1998;13(1):101-8.

[51] Kumar SKN, Kumar RM, Thanushkodi K, Renuga P. Reactive power planning considering the highest load buses using evolutionary programming. Int J Recent Trends Eng 2009;2(6):37-9.

[52] Padaiyatchi S, Daniel M. OPF-based reactive power planning and voltage stability limit improvement under single line outage contingency condition through evolutionary algorithms. Turk J Electr Eng Comput Sci 2013;21(4):1092-106.

[53] Gopalakrishnan V, Thirunavukkarasu P, Prasanna R. Reactive power planning using hybrid evolutionary programming method. In: Proceedings of IEEE/PES power systems conference and exposition, New York; 2004. p. 1319-23.

[54] Musirin I, Abdul Rahman TK. Evolutionary programming optimization technique for solving reactive power planning in power system. In: Proceedings of the 6th WSEAS international conference on evolutionary computing, Lisbon, Portugal; 2005. p. 239-44.

[55] Sheng S, Li J. Study of reactive power optimization based on artificial immune ant colony algorithm. In: Proceedings of the 3rd international conference on electric utility deregulation and restructuring and power technologies - DRPT, Nanjuing, China; 2008. p. 2311-5.

[56] Rajkumar P, Devaraj D. Particle swarm optimization approach for constrained reactive power planning. In: Proceedings of international conference in power electronic drives and power system, Sona College of Technology, Salem; 2008. p. 1-7.

[57] Krami N, El-Sharkawi MA, Akherraz M. Pareto multiobjective optimization technique for reactive power planning. In: Proceedings of IEEE power and energy society general meeting -conversion and delivery of electrical energy in the 21st century, Pittsburgh; 2008. p. 1-6.

[58] Sutha S, Kamaraj N. Optimal location of multi type facts devices for multiple contingencies using particle swarm optimization. World Acad Sci, Eng Technol 2008;22:791-7.

[59] De M, Goswami SK. Optimal reactive power procurement with voltage stability consideration in deregulated power system. IEEE Trans Power Syst 2014;29(5):2078-86.

[60] Bhattacharyya B, Kumar S. Reactive power planning with FACTS devices using gravitational search algorithm. Ain Shams Eng J 2015;6:865-71.

[61] Shaw B, Mukherjee V, Ghoshal SP. Solution of reactive power dispatch of power systems by an opposition-based gravitational search algorithm. Int J Electr Power Energy Syst 2014;55:29-40.

[62] Rajan A, Malakar T. Optimal reactive power dispatch using hybrid Nelder-Mead simplex based firefly algorithm. Int J Electr Power Energy Syst 2015;66:9-24.

[63] Ghasemi M, Ghanbarian MM, Ghavidel S, Rahmani S, Moghad-dam EM. Modified teaching learning algorithm and double differential evolution algorithm for optimal reactive power dispatch problem: a comparative study. Inf Sci 2014;278:231-49.

[64] Dutta S, Roy PK, Nandi D. Optimal location of STATCOM using chemical reaction optimization for reactive power dispatch problem. Ain Shams Eng J 2015.

[65] Lenin K, Reddy BR, Kalavathi MS. Water cycle algorithm for solving optimal reactive power dispatch problem. J Eng Technol Res 2014;2(2):1-11.

[66] Amrane Y, Boudour M, Belazzoug M. A new optimal reactive power planning based on differential search algorithm. Int J Electr Power Energy Syst 2015;64:551-61.

[67] Zhang W, Liu Y. Multi-objective reactive power and voltage control based on fuzzy optimization strategy and fuzzy adaptive particle swarm. Int J Electr Power Energy Syst 2008;30(9):525-32.

[68] Rajkumar P, Devaraj D. A Hybrid particle swarm optimization approach for reactive power optimization in electricity market. Int J Energy Technol Policy 2011;7(4):379-92.

[69] Kheirizad I, Mohammadi A, Varahram MH. A novel algorithm for optimal location of FACTS devices in power system planning. J Electr Eng Technol 2008;3(2):177-83.

[70] Subbaraj P, Rajnarayanan PN. Hybrid particle swarm optimization based optimal reactive power dispatch. Int J Comput Appl 2010;1(5):65-70.

[71] Keko H, Duque AJ, Miranda V. A multiple scenario security constrained reactive power planning tool using EPSO. Int J Eng Intell Syst Electr Eng Commun 2007;15(2):83-90.

[72] Abou El-Ela AA, El Sehiemy R, Shaheen AM. Multi-objective fuzzy-based procedure for enhancing reactive power management. IET Gener Transm Distrib 2013;7(12):1453-60.

[73] Bhattacharyya B, Goswami SK, Bansal RC. Hybrid fuzzy particle swarm optimization approach for reactive power optimization. Int J Electr Syst 2009;5(3):1-15.

[74] El Sehiemy R, Abou El Ela AA, Shaheen AM. A multi-objective fuzzy-based procedure for reactive power-based preventive emergency strategy. Int J Eng Res Afr 2015;13:91-102, http:// www.scientific.net/.

[75] Ambika R, Rajeswari R, Nivedita A. Comparative analysis of nature inspired algorithms applied to reactive power planning studies. Indian J Sci Technol 2015;8(5):445-53.

[76] Jeyadevi S. Evolutionary algorithms based solution to reactive power planning and dispatch problems with security constraints. Ph.D. thesis. Faculty of Electrical Engineering, Anna University, Chennai; December 2010.

[77] Jeyadevi S, Baskar S, Willjuice M. Reactive power planning with voltage stability enhancement using covariance matrix adapted evolution strategy. Eur Trans Electr Power 2011;21(3):1343-60.

[78] Binitha S, Sathya SS. A survey of bio inspired optimization algorithms. Int J Soft Comput Eng (IJSCE) 2012;2(2):137-51.

[79] General algebraic modeling system (GAMS). http://www. gams.com.

[80] Roselyn JP, Devaraj D, Dash SS. Multi objective differential evolution approach for voltage stability constrained reactive power planning problem. Int J Electr Power Energy Syst 2014;59:155-65.

Dr. Shimaa Rabah Spea is a lecturer at electrical engineering department, Faculty of Engineering, Menoufiya University, Egypt. She received her B.Sc. at Shebin El-kom Faculty of Engineering -Menoufiya University in 2003 while she received her M.Sc. and PHD Degrees in 2006 and 2010 at Shebin El kom Faculty of Engineering - Menoufiya University, respectively. Her research interests include Power system operation, control, and planning, applications of optimization algorithms in electric power systems.

Prof. Sobhy M. Farrag was born in Shebin El-kom, Egypt in 1950. He received his B.Sc. degree from EL-Menoufiya University in 1973, his M.Sc. degree from El-Mansoura University 1978, and his Ph.D. degree from Slovak Technical University, Bratislava, Slovakia, in 1983. His research interests include power system planning, economic operation of electric power systems, renewable energy sources, distrusted generation (DG), reactive power planning and voltage control. He is now a professor at the Electrical Engineering Department, El-Menoufiya University, Shebin El-kom, Egypt.

Prof. Mohammed A. Abido is a professor at electrical Engineering Department, King Fahd University of Petroleum and Minerals, Saudi Arabia. He received his B.Sc. and M.Sc. at Shebin El-kom Faculty of Engineering -Menoufiya University in 1985 and 1989, respectively. He received his PHD Degree in 1997 at King Fahd University of Petroleum & Minerals (KFUPM), Saudi Arabia. His research interests include power system operation and control, flexible ac transmission systems (FACTS), evolutionary algorithms,neuro-fuzzy control and hybrid intelligent systems, reactive power control and voltage stability, and renewable energy.

Eng. Abdullah Mohammed Shaheen is an

electric power engineer at South Delta Electricity Distribution Company (SDEDCo), Ministry of Electricity, Tanta, Egypt. He received his B.Sc. 2007 at Port Said Faculty of Engineering - Suez Canal University in 2007 while he received his M. Sc. Degree in 2012 at Shebin El kom Faculty of Engineering - Menoufiya University in 2012. His research interests comprise Power system operation, control, and planning, applications of optimization algorithms in electric power systems, renewable sources, and smart grid.