A Literature Survey of Optimal Power Flow Problems in the Electricity Market Context

Zhifeng Qiu, Graduate Student Member, IEEE, Geert Deconinck, Senior Member, IEEE, Ronnie Belmans, Fellow, IEEE

Abstract--For decades researchers have developed various models and algorithms to look for the Optimal Power Flow (OPF) in different applications. Still research is ongoing to find OPF problems for the present day power system challenges such as a liberalized market or a large penetration of renewable energy source. Traditionally, classical mathematical optimization methods have been used to effectively solve conventional OPF problems. Due to emergence of a deregulated electricity market and consideration of dynamic system properties, however, the traditional concepts and practices of power systems are overruled by an economic market management. So the requirements for OPF have become more complex than it was. The purpose of this survey is to collect information from the previous research literatures and to classify different modeling and solving approach for OPF. Index TermsOptimal Power Flow, Electricity Market.

I. INTRODUCTION

HE optimal power flow (OPF) problem has been one of the most widely studied subjects in the power system community since Carpentier first published the concept in 1962 [1]. The objective of an Optimal Power Flow (OPF) algorithm is to find a steady state operation point which minimizes generation cost, network loss etc. or maximizes social welfare or system utilization etc. while maintaining an acceptable system performance in terms of limits on generators active and reactive powers, line flow limits, maximum output of various compensating devices etc.. Take optimal active power dispatch as an example, the objective function is to minimize total generation cost or network losses while the system is operating within its security limits. After computing, OPF gives the optimal active power generated and purchased at each bus and the nodal prices which are important conception in the electricity market. Nodal prices [2] are of specific interest because they reflect the marginal cost of delivering one more unit at each bus (node). These prices are also called locational prices and are found to be the optimal prices, maximizing social welfare and taking transmission constraints into account. In an uncongested grid all locational price are equal, whereas in case of congestion locational price difference occurs. They can provide the right incentives to market players and to society. OPF algorithms are among the tools present in

many Energy Management Systems (EMSs) and their usefulness is increasingly being recognized by power utilities due to increased presence of independent power producers combined with deregulation of the power industry. The traditional OPF had no such rich techno- economic meaning as new form OPF explicitly has. Research result of OPF in deregulated electricity market context can be extended into many research areas: locational real-time pricing, network congestion management, available transfer capability estimation, electricity transmission fee computation, etc. An extensive OPF survey can provide input to further research. Aiming at various application domains, implementations of OPF differ in terms of the model formulation, the level of detail in the representation of the system and the computational/algorithmic aspects of the solution methodology. Different objectives call on different levels of detail in the model and in the representation of various considerations. However, there is considerable arbitrariness involved in all aspects of the OPFmodel formulation, level of detail in the interconnected system representation and solution methodology/ implementation scheme. The following table gives a simple illustration of OPF application in power system.
TABLE I DIFFERENT APPLICATIONS OF OPF IN POWER SYSTEMS Application Field Real-time electricity price computing Network congestion management Electricity transmission fee allocation Available transmission capability (ATC) computing Objective Functions of OPF Model Max. social welfare Min. congestion management Min. generation cost/ max custom benefits Max total transmission capability

This paper is organized as follows: Section two, the different Optimal Power Flow problem with different objective function and different constraints is summarized. Section three classifies the different optimization method to solve OPF problems. Section four gives the conclusion. II. FORMULATION OF OPF OPF is formulated mathematically as a general constrained optimization problem.

978-1-4244-3811-2/09/$25.00 2009 IEEE

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.

Min F (u , x) s.t. h(u, x) = 0

(1.a ) (1.b)

A. Objective function The objective function of the OPF problem could be varied based on the target of the optimization. Normally, the objective function has the following forms: a> minimal system operation cost (cost of active generation) b> minimal active power transmission loss c> minimal exhaust emission d> minimal ancillary service cost e> maximal load dispatch capacity a> and b> are two forms often used. Take a> for example, the costs may be defined as polynomials or as piecewise-linear functions of generator output. Besides the single optimal problem with only one optimization goal, in a number of real-world problems, the decision maker is faced with multiple goals. The presence of multiple objectives in a problem, in principle, gives rise to a set of optimal solutions known as Pareto-optimal solution, instead of a single optimal solution. The multi-objective optimization (MO) techniques in power system [3] have the following advantages: (a) it allows the management of different objectives; (b) it simplifies the decision making process and (c) it gives indications on the consequences of the decision with respect to all the objective functions considered. There are several methods to solve such kind problems [4-10]. One is that the multi-objective OPF problem was converted to a single objective problem by a linear combination of different objectives as a weighted sum [4]. Unfortunately, this requires multiple runs as many times as the number of desired Pareto-optimal solutions. Furthermore, this method cannot be

B. Constraints Conventional constraints include equality constraints (active power balance equations, reactive power balance equations) inequality constraints (apparent power flow limit of line from and to side, bus voltage limits, active and reactive power generation limits) [11]. Such constraints belong to steady-state constraints without considering system contingencies that can occur temporarily. They are easier to work with than those considering transient-stability constrained OPF and OPF with FACTS (flexible AC transmission systems). Transient-stability constrained OPF(TSCOPF) One of the deregulated market goals of any electric power utility is to provide reliable and uninterrupted service to its customers. There is need to include suitable security constraints within the whole market pricing mechanisms, so that correct market signals can be sent to all market participants while operating the system within reasonable security margins. The report on the August 14, 2003 blackout in Canada and U.S. [13] has pointed out that after the outage of a 345 kV transmission line the power system has moved to an operating condition where it was unable to sustain additional contingencies. The report highlighted the importance of implementing suitable corrective actions to return the power system within its operating security limit. Transient-stability constrained OPF can be mathematically expressed with inequality constraints imposed by the rotor angle limits [12]. The power flow solution should not only meet the steady-state constraints imposed by the conventional OPF problem but also the dynamic constraints imposed on the rotor angles during the transient period under study for a given set of contingencies. TSCOPF is a nonlinear optimization problem with both algebraic and differential equations, which is difficult to be solved even for small power systems. The main difficulties of solving this problem include [14]: (1) how to deal with the differential equations that represent the dynamic behavior of the system; and (2) how to deal with the disruption of

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.

g (u , x) 0 (1.c) Where, u is the set of controllable variables in the system and x is the set of dependent variables. For example, u consists of generator voltage, generator active power output except at the slack bus, transformer tap settings and shunt VAR compensations; x consists of slack bus power, load bus voltages, generator reactive power outputs and transmission line loadings. F(u,x) is an objective function which is a scalar. Equality constraints (1.a) are derived from conventional power balance equations. Inequality constraints (1.c) are the limits on control variables u and the operating limit on the other variables of the system. The non-linear programming problem (1) has been solved by a wide range of numerical techniques. An important byproduct of the optimum of the OPF is the economic information obtained from the dual variables, which has direct applications in electricity markets [1]. For example, the dual variables associated with the power flow equality constraints can be interpreted as nodal active and reactive power prices at each bus. Similarly, dual variables associated with inequality constraints are interpreted as the sensitivity of the objective function to a change in the constraint limit.

used to find Pareto-optimal solutions in problems having a non-convex Pareto-optimal front. Another method is to use a fuzzy set to define the different membership functions and degrees of each objective function and of the scalable variables [5]. The objective of multiobjective OPF is transferred into finding the minimal i from all the membership degrees and making this minimal i as large as possible by controlling the system variables. Hence, Max = min {1 , 2 , , n } . Studies on evolutionary algorithms, over the past few years, have shown that these methods can be efficiently used to eliminate most of the difficulties of classical methods [6, 7]. Since they use a population of solutions in their search, multiple Pareto-optimal solutions can be found in one single run. The multi-objective evolutionary algorithms have been implemented to different power system optimization problem with impressive success [8-10].

conventional optimization algorithms in highly nonlinear and nonconvex solution landscapes. OPF with FACTS The use of FACTS devices to stretch transmission capacity makes it possible that the power system is operated at the minimal cost while satisfying specified transmission constraints and security constraints [15]. In order to model FACTS devices in OPF, the conventional OPF program must undergo some changes such as inclusion of new control variables belonging to FACTS devices and the corresponding load flow solutions to deal with the above said problem [16]. Issues include increased utilization of existing infrastructure such as secure system operation at higher power transfers across existing transmission lines which are limited by stability constraints, the development of control designs for FACTS devices, and determination of functional performance requirements for FACTS components. Different FACTS devices, such as static var compensators (SVC), solid state synchronous compensators (SSSC), thyristor controlled series capacitors (TCSC), unified power flow controller (UPFC) etc, show different behavior in the power system. Existing algorithms are usually confined to specific power systems cases that do not offer great freedom in objective functions or the types of constraints that may be used. Hence it is important to invent more general and reliable algorithms, which are capable of dealing new constraints arising from FACTS devices. III. ALGORITHMS TO SOLVE OPF This section reviews various optimization methods used to solve OPF problems. The earliest OPF algorithms started from classical mathematical programming methods. Linear programming (LP), quadratic programming (QP), nonlinear programming (NLP), and interior point methods (IPM) have successfully shown powerful capabilities in this field in the last decades. J.A.Momoh [43] presented a literature review of such methods which had appeared up to 1993. Due to the weakness of the algorithm itself, still some problem exists, such as convergence. Comparably, another set of promising methods, artificial intelligence (AI) plays a more and more important role in OPF field. AI methods such as genetic algorithms (GA), evolutionary programming (EP), and particle swarm optimal (PSO) have been employed to overcome some drawback of conventional techniques. A. Linear Programming (LP) Method The linear programming formulation requires linearization of the objective function and constraints with nonnegative variables. In general, this method firstly executes piecewise linearization or successive linearization to approximate the nonlinear function, and then use the linear programming approach. In 1968, Wells [22] first proposed using LP to solve economic dispatch with security constraints. The idea is that after linearization of objective function and constraints,

simplex method is applied to solve the optimality problem. This method has two drawbacks: 1> in unfeasible conditions, the final result is not optimal solution. 2> constraints will be overloaded due to computing rounding errors. Lima et al. [23] made use of mixed integer linear programming to conduct a design study on the combinatorial optimal placement of Thyristor Controlled Phase Shifter Transformers (TCPSTs) in large-scale power systems. It studied the number, network location and settings of phase shifters that maximize system loadability under the DC load flow model, subject to limits on the installation investment or total number of TCPSTs are studied. Computation time is significantly lower than those of other published comparable cases. B. Nonlinear Programming (NLP) Method A nonlinear model is the earliest OPF mathematical formulation. Nonlinear programming (NLP) deals with optimality problems with nonlinear objective and/or constraint functions. NLP is divided into two kinds, constraint NLP and unconstraint NLP. The Lagrangian method or penalty function method can be used to transfer constraint NLP into unconstraint NLP. Many algebraic programming methods can be employed for this kind of unconstraint NLP. In 1968, Dommel and Tinney[17] are the first that successfully used a simplified gradient descent algorithm to solve OPF. This algorithm is based on the Newton method. The independent variables are the control variables of system. The inequality constraints are dealt by assigning penalty functions and judging boundary violations using the Lagrangian multiplier method. However, sometimes convergence performance becomes worse due to this penalty function method. In the following years, a lot of research [18-20] was done to improve NLP for OPF. For example, Jizhong Zhu and Momoh [21] proposed a nonlinear convex network flow programming (NLCNFP) model and algorithm for solving the security constrained multi-area economic dispatch (MAED) problem. In MAED, the tie-line security and transfer constraints in each area are considered. A buying and selling contract in a multiarea environment is analyzed. NLCNFP is solved by using a combined method of quadratic programming and network flow programming. Their test results on four interconnected power systems show that the method is feasible and effective. C. Quadratic Programming (QP) Method QP is a special form of nonlinear programming, whose objective function must be quadratic and constraints must be linear. In 1973, Reid and Hasdorf [24] first proposed using QP to solve economic dispatch problem. The algorithm uses help variables to approximate cost functions into quadratic functions. Constraints are linearized by using Taylor series expansion. Then optimal solution is obtained by Wolfe algorithm. The convergence performance is not affected by

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.

selection of step length and penalty factor. However, the computation time will become longer with larger system scale. In 1982, a big breakthrough was made by Burchett et.al [25]. In their work, the original nonlinear OPF problem has been decomposed into sequential quadratic programming problem. Compared with the old algorithms, the augmented Lagrangian method demonstrates the advantages of reasonable run times and robustness. Momoh [26] presented an extension of basic Kuhn-Tucker conditions to employ a generalized quadratic-based model for OPF. The conditions for feasibility, convergence and optimality are included in the construction of the OPF algorithm. It is also capable of using hierarchical structures to include multiple objective functions and selectable constraints. The generalized algorithm using sensitivity of objective functions with optimal adjustments in the constraints yields a global optimal solution. Computational memory and execution time required have been reduced with respect to other approaches.

interior point method with a step control procedure to enhance the convergence of market-based OPF computation. TRALM is more theoretically rigid, but in terms of computational performance, SCIPM is better for real-time applications.

E. Artificial Intelligence (AI) Methods The classical optimization methods mentioned above are limited in handling the dynamic characteristics such as the transient stability performance in the optimization and may suffer from the high sensitivity to initial conditions. In order to overcome these weaknesses, heuristic optimization techniques, such as evolutionary algorithms (EA), genetic algorithm (GA), particle swarm optimization (PSO), etc. have been developed. Genetic Algorithm (GA) Method Genetic Algorithms belong to the category of random search algorithms which simulate the evolution process based on the theory of survival of the fittest. It is a particular class of evolutionary algorithms which can be categorized as global search heuristics. Po-H.Chen et al. T.S.Chung et al. [32] presented a Hybrid Genetic Algorithm (GA) method to solve OPF incorporating D. Interior Point (IP) Method Among earlier techniques applied to solve the OPF problem, FACTS devices. GA is integrated with conventional OPF to the Interior Point method is considered one of the most select the best control parameters to minimize the total efficient, due to its good performance and robustness in the generation fuel cost and keep the power flows within the solution of linear and nonlinear programming problems. The security limits. TCPS and TCSC are modeled. The proposed main idea of this algorithm is to search for an optimal point in method was applied on modified IEEE 14 bus system and it a feasible region. converged in a few iterations. In 1954, Frish [27] proposed the earliest IP method. It is a Reference [33] presents an efficient real-coded mixedkind of the barrier function method to solve unconstraint integer genetic algorithm (MIGA) for solving non-convex optimality problem. optimal power flow (OPF) problems while considering With the development of linear algebra and increased transmission security and bus voltage constraints for practical computing performance, IP method progressed much. application. In the MIGA method, the individual is the realKarmarkar [28] proposed a new method in 1984 for solving coded representation that contains a mixture of continuous and large-scale linear programming problems very efficiently. It is discrete control variables, and two arithmetic crossover and known as an interior method since it finds improved search mutation schemes are proposed to deal with directions strictly in the interior of the feasible space. The continuous/discrete control variables, respectively. The computing speed is 50 times as fast as the simplex method. experimental results show that the MIGA-based OPF method is Afterwards, Gill [29] generalized IP into nonlinear superior to the Evolutionary Programming. programming. In [34], Bakirtzis presents an enhanced genetic algorithm Nowadays IP methods are a popular and efficient algorithm with both continuous and discrete control variables. The for solving OPF problem. There are many improved IP method. continuous control variables modeled are unit active power Wei Yan et al. [30] presented the solution of the Optimal outputs and generator-bus voltage magnitudes, while the Reactive Power Flow (ORPF) problem by the Predictor discrete ones are transformer-tap settings and switchable shunt Corrector Primal Dual Interior Point Method (PCPDIPM). A devices. Advanced problem-specific operators are introduced new optimal reactive power flow model in rectangular form is in order to enhance the algorithms efficiency and accuracy. proposed. The form is quadratic, which allows for ease of Numerical results on two test systems are presented and matrix setup and inexpensive incorporation of higher order compared with results of other approaches. Evolutionary Programming (EP) information in a predictor-corrector procedure that reduces the Evolutionary Programming is stochastic global search number of IPM iterations to convergence. The total calculation time needed for the proposed method is always shorter than methods that mimic the metaphor of natural biological evolution, and have been successfully applied to many power that for the conventional model for other test cases. In [31], Hongye Wang et al. proposed a step-controlled system optimization problems. Reference [35] proposed using EP for market bidding primal-dual interior point method (SCIPM) for reliable and efficient computation of large-scale market-based OPFs. strategies for a day-ahead energy and reserve. The optimal Compared with trust-region based augmented Lagrangian bidding parameters for both markets are determined by solving method (TRALM), SCIPM amends the popular primal-dual an optimization problem that takes unit commitment

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.

constraints such as generating limits and unit minimum up/down time constraints into account. The proposed algorithm is developed from the view point of a generation company wishing to maximize a profit as a participant in the deregulated power and reserve markets. Separate power and reserve markets are considered, both are operated by clearing price auction system. Reference [36] aims at reviewing the performance of the different evolutionary programming (EP) techniques for all kinds of economic dispatch (ED) problems. The three EP techniques considered here differ in the kind of mutation they use: Gaussian, Cauchy and combined GaussianCauchy mutation. The kinds of problems tested are: ED of generators with prohibited operating zones (POZ), ED of generators with piecewise quadratic cost functions (PQCF), combined economicenvironmental dispatch (CEED) and multi-area economic dispatch (MAED). Numerical examples typical to each kind of ED have been tested. The simulation results are compared and discussed to show the relative performances of the different EP techniques. Within EA, a new branch, differential evolution (DE) developed by Storn and Price [37] has gained attention recently due to its strong ability in searching a global optimal solution. Studies have shown that DE is simpler, more efficient and robust than other EAs in many benchmark problems [38]. Reference [14] develops DE for solving transient stability constrained OPF. Due to the flexibility of DE, the hybrid method combines time-domain simulation with a transient energy function method is employed in DE. Several strategies are proposed to reduce the computational burden, such as the initialization, assessment and selection of solution individuals in evolution process of DE. The effectiveness of the DE is shown by test on small and large-scale system. Particle Swarm Optimization (PSO) Particle swarm optimization (PSO) is a kind of stochastic, population-based optimization algorithm, which was inspired by the social behaviors of animals such as fish schooling and bird flocking. It is a kind of swarm intelligence. It was introduced by Kennedy and Eberhart in 1995 as an alternative to genetic algorithms (GA) [39, 40]. It requires only few parameters to be tuned and hence is attractive from an implementation viewpoint. The first version of PSO was intended to handle only nonlinear continuous optimization problems. However, its development elevated its capabilities for handling a wide class of complex optimization problems. Reference [41] employs the particle swarm optimization (PSO) algorithm with reconstruction operators (PSO-RO) as the optimization tool to solve security constraints OPF with continuous and discrete control variables. The operators guarantee searching the optimal solution within the feasible space, reducing the computation time and improving the quality of the solution. The proposed approach is validated on systems from the specialized literature Jong-Bae Park et al. [42] proposed a modified PSO (MPSO) for economic dispatch with nonsmooth cost functions

considering valve-point effects and multi-fuel problems. A position adjustment strategy is proposed to provide the solutions satisfying the inequality constraints. The equality constraint is resolved by reducing the degree of freedom by one at random. Dynamic search-space reduction strategy is devised to accelerate the process. Results of MPSO compared its superiority to other method. IV. CONCLUSION This paper reviews the OPF problem from modeling method to optimal algorithm in the context of electricity markets. Its obvious from the existing research literature work that, OPF has a large application field in power system and also has achieved several successes in aspects application in real-time pricing or congestion management. For the OPF modeling aspect, existing OPF models still need to be enhanced to be more practical and more complete models in order to depict real dynamic power system. Dynamic constraints of system stability, constraints and additional variables brought by FACTS devices and large-scale hybrid system (including several types power generators) should be considered more in the future research. For the algorithm aspects, conventional mathematical methods suffer from poor convergence, are easily stuck in a local optimum, find only one solution in one computation run, are slow and have limited capabilities for large-scale system. Many methods have been proposed to improve these drawbacks. AI methods have the advantage that they are versatile for handling various qualitative constraints, suitable for multi-objective optimal problems, and converge rapidly. However, some AI methods have the disadvantage of a longer computation time, finding local optimum, or no physical meaning. So it is wise to combine these two methods, which allows the disadvantage of each method to be remedied with the help of the other methods advantage. In electricity market, when algorithms employed to get the economic values, i.e. nodal price, mathematical methods have advantages which AI methods havent. That is mathematical methods can provide the economic information (shadow price) indicated by the Lagrangian multipliers. Comparably, the AI methods just give the solution of system including optimal value of objective function and control variables without the economic byproduct. V. REFERENCES
[1] [2] [3] [4] [5] J. Carpentier, Contribution a lEtude du Dispatching Economique, Bull. Soc. Francaise des Electriciens, vol. 3, pp. 431447, Aug. 1962. D. Ray, F. Alvarado. Use of an engineering model for economic analysis in the electricity utility industry. presented at the Advanced Workshop on Regulation and Public Utility Economics, May 25-27, 1988 Kralj B., Rajakovic N.: Multi-objective programming in power system optimization: new approach to generator maintenance scheduling, Int. J. Elect. Power Energy Syst., 16, (4), pp. 211220, 1994, M. A. Abido, Multiobjective particle swarm optimization for optimal power flow problem, Power System Conference, MEPCON 2008. 12th International Middle-East, 12-15 March, 2008 Xiaohong Guan, W.H. Liu, Alex D. Papalexopoulos. Application of a fuzzy set method in an optimal power flower. Electric Power Systems Research, Vol.34, pp. 11-18, 1995

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

J. Kennedy and R. Eberhart, Swarm Intelligence, Morgan Kaufmann Publishers, 2001. M. A. Abido, Multiobjective Evolutionary Algorithms for Electric Power Dispatch Problem IEEE Trans. on Evolutionary Computations, Vol. 10, No. 3, pp. 315-329, June 2006. M. A. Abido, Environmental/Economic Power Dispatch Using Multiobjective Evolutionary Algorithms IEEE Trans on Power Systems, Vol. 18, No. 4, pp. 1529-1537, November 2003. J. B. Park, K. S. Lee, J. R. Shin, and K. Y. Lee, A Particle Swarm Optimization for Economic Dispatch With Nonsmooth Cost Functions, IEEE Trans. on Power Systems, Vol. 20, No. 1, pp. 34-42, 2005. A. I. Selvakumar and K. Thanushkodi, A New particle Swarm Optimization Solution to Nonconvex Economic Dispatch Problems, IEEE Trans. on Power Systems, Vol. 22, No. 1, pp. 42-51, 2007 Xifan Wang. Modern electric power system analysis, Beijing,2003 D.Gan, R. Thomas, R. Zimmerman Stability-Constrained Optimal Power Flow, IEEE Trans. on Power Systems, Vol.15, No.2, May 2000, pp. 535-540. Canada-U.S. Power System Outage Task Force, Interim Report: Causes of the August 14th Blackout in the United States and Canada, Natural Resources Canada, November 2003. H. R. Cai, C. Y. Chung, and K. P. Wong, Application of Differential Evolution Algorithm for Transient Stability Constrained Optimal Power Flow, IEEE Trans. on Power Systems, VOL. 23, NO. 2, MAY 2008 Hammons TJ., Lim S.K. Flexible AC Transmission System(FACTS). Electric Machines & Power System Vol.25, pp. 73-851997 Abdel-Moamen M. A Narayana Prasad Padhy, Optimal Power Flow Incorporating FACTS Devices- Bibliography and Survey, Transmission and Distribution, Conference and Exposition, 2003 IEEE PES, Vol, 2, pp, 669- 676 2003 Dommel, H.W. Tinney, W.F. Optimal Power Flow Solutions, IEEE Transactions on Power Apparatus and Systems, Vol: PAS-87, pp: 18661876, 1968 N. Grudinin, Reactive power optimization using successive quadratic programming method, IEEE Trans. Power Syst., vol. 13, no. 4, pp. 1219-1225, Nov. 1998. A. Berizzi, M. Delfanti, P. Marannino, M. S. Pasquadibisceglie and A. Silvestri, Enhanced security-constrained OPF with FACTS devices, IEEE Trans. Power Syst., vol. 20, no. 3, pp. 1597-1605, Aug. 2005. S.N. Talukdar, T.C.Giras, et. Decomposition for optimal power flow. IEEE Trans. Power Apparatus and Systems, Vol: PAS-102, pp: 38773884, 1983 J. A. Momoh and J. Zhu, Multi-area power systems economic dispatch using nonlinear convex network flow programming, Electric Power Syst. Research, vol. 59, pp. 13-20, 2001. D.W.Wells. Method for economic secure loading of a power systems. Proceedings of IEEE, Vol, 115, pp, 606- 614 , 1968 F. G. M. Lima, F. D. Galiana, I. Kockar and J. Munoz, Phase shifter placement in largescale systems via mixed integer linear programming, IEEE Trans. Power Syst, vol. 18, no. 3, pp. 1029-1034, Aug. 2003. G.F.Reid, L. Hasdorf. Economicdispatch using quadratic programming. IEEE Transactions on Power Apparatus and Systems, Vol: PAS-92, pp: 2015-2023, 1973 Burchett R C, H. H. Happ, Vierath D R, Quadratic ally Convergent Optimal Power Flow. IEEE Transactions on Power Apparatus and Systems. Vol: PAS-103, pp: 3267-3276, 1984 J. A. Momoh, A generalized quadratic-based model for optimal power flow, Systems, Man and Cybernetics, 1989. Conference Proceedings., IEEE International Conference on , pp. 261- 267.1989 K.R. Frish. Principles of linear programming- the double gradient form of the logarithmic potential method. Memorandum, Institute of economics, University of Oslo, Oslo, Norway, 1954 N. Karmarkar. A new polynomial-time algorithm for linear programming. Combinatorica, Vol. 4, pp. 3737-395, 1967 P.E.Gill, W.Murray, et.al. On the progected Newton barrier methods for linear programming and an equivalence to Karmarkars projective method. Math. Programming, Vol. 36, pp. 183-209, 1986 Wei Yan, J. Yu, D. C. Yu and K. Bhattarai, A new optimal reactive power flow model in rectangular form and its solution by predictor corrector primal dual interior point method, IEEE Trans. Power Syst., vol. 21, no. 1, pp. 61-67, Feb. 2006.

[31] H. Wang, C. E. Murillo-Snchez, R. D. Zimmerman, R. J. Thomas, On Computational Issues of Market-Based Optimal Power Flow, IEEE Transactions on Power Systems, Vol. 22, No. 3, Aug. 2007, pp. 11851193. [32] T. S. Chung and Y. Z. Li, A hybrid GA approach for OPF with consideration of FACTS devices, IEEE Power Engineering Review, pp. 47-50, Feb. 2001. [33] Zwe-Lee Gaing, Rung-Fang Chang, Security-Constrained Optimal Power Flow by Mixed-Integer, Power Engineering Society General Meeting, 2006. IEEE [34] Bakirtzis, A. G. Biskas, P. N. Zoumas, C. E. Petridis, V. Optimal Power Flow by Enhanced Genetic Algorithm, Power Engineering Review, IEEE, 2002 [35] P. Attaviriyanupap, H. Kita, E. Tanaka, J. Hasegawa, New bidding strategy formulation for day-ahead energy and reserve markets based on evolutionary programming, Electrical Power and Energy Syst., vol. 27 pp.157-167, 2005. [36] T. Jayabarathi, K. Jayaprakash, D. N. Jeyakumar and T. Raghunathan, Evolutionary programming techniques for different kinds of economic dispatch problems, Electric Power Syst. Research, 73, pp.169-176, 2005. [37] R. Storn and K. Price, Differential evolution-a simple and efficient adaptive scheme for global optimization over continuous spaces, Int. Comput. Sci. Inst., Berkeley, CA, Tech. Rep., 1995. [38] J. Vesterstrom and R. Thomsen, A comparative study of differential evolution, particle swarm optimization, and evolutionary algorithms on numerical benchmark problems, in Proc. Congr. Evolutionary Computation, Portland, OR, Jun. 1923, 2004, vol. 2, pp. 19801987. [39] Kennedy I. and Eberhart R. C., Particle swarm optimization, Proceedings of IEEE International Conference on Neural Networks, Piscataway, NJ. pp. 1942-1948, 1995. [40] Eberhart R. C. and Kennedy J., A new optimizer using particle swarm theory, Proceedings of the Sixth International Symposium on Micromachine and Human Science, Nagoya, Japan, pp. 39-43, 1995. [41] Pablo E. Oate Yumbla, Juan M. Ramirez, , Carlos A. Coello Coello, Optimal Power Flow Subject to Security Constraints Solved With a Particle Swarm Optimizer, IEEE Trans. Power Syst., VOL. 23, NO. 1, FEBRUARY 2008 [42] J. B. Park, Ki. S. Lee, J. R. Shi and K. Y. Lee, A particle swarm optimization for economic dispatch with nonsmooth cost functions, IEEE Trans. Power Syst., vol. 20, no. 1, pp. 34-42, Feb. 2005. [43] J. A. Momoh, M. E. El-Harwary and Ramababu Adapa, A review of selected optimal power flow literature to 1993, part I and II, IEEE Trans. Power Syst., vol. 14, no. 1, pp. 96-111, Feb. 1999. Qiu Zhifeng received the M.Sc. degree in automatic control engineering from the Central South University (CSU), Changsha, China, in 2004. She worked as a teach assistant in Information Science and Technology School of CSU from 2004.9 to 2006.12. Now she is a Ph.D. candidate in Katholieke Universiteit Leuven (KUL). Geert Deconinck (S88M96SM00) received the M.Sc. degree in electrical engineering and the Ph.D. degree in applied sciences from the Katholieke Universiteit Leuven (KUL), Leuven, Belgium, in 1991 and 1996, respectively. Since 1997, he has been with the staff of the Department of Electrical Engineering, KUL, first as a Postdoctoral Fellow and now as an Associate Professor. His research interests include the design, analysis, and assessment of software-based fault-tolerant solutions to meet real-time, dependability, and cost constraints for embedded applications on parallel and distributed systems. Ronnie Belmans (S77M84Fellow02) received the M.Sc. and Ph.D. degrees in electrical engineering from the Katholieke Universiteit Leuven (KUL), Leuven, Belgium, in 1979 and 1984, respectively. He is currently a Full Professor in the Department of Electrical Engineering, KUL, where he teaches electrical energy systems including electrical machines. He also serves as Chairman of the Board of Directors at the Belgian Transmission System Operator Elia. He has been a Visiting Professor at McMaster University, Hamilton, ON, Canada, Rheinisch-Westflische Technische Hochschule Aachen, Aachen, Germany, and the Imperial College of Science, Technology and Medicine, London, U.K. His research interests include power quality, electrical energy systems, and electrical machine design.

Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on June 25,2010 at 23:19:46 UTC from IEEE Xplore. Restrictions apply.