C I R E D 19 th International Conference on Electricity Distribution Vienna, 21-24 May 2007 Paper 0088 CIRED2007 Session 5 Paper No 0088 Page 1 / 4 ENHANCED PARTICLE SWARM OPTIMIZATION METHOD FOR POWER LOSS REDUCTION IN DISTRIBUTION SYSTEMS Mihai GAVRILAS Ovidiu IVANOV Calin Viorel SFINTES Technical University, Iasi – Romania Technical University, Iasi – Romania Eximprod S.A. - Romania mgavril@ee.tuiasi.ro ovidiuivanov@ee.tuiasi.ro calin.sfintes@eximprod.ro ABSTRACT This paper presents a new approach to the reactive power compensation in distribution systems using an enhanced version of the Particle Swarm Optimization algorithm. Second order correction terms are used to compute the velocity of each particle during the search process. The new approach has better convergence properties. INTRODUCTION One of the major concerns of a power distribution company (DISCO) is to reduce technical and non-technical losses, while ensuring that the demand is satisfied at any moment at a reliable level of system performance and with the lowest possible cost. At present, due to the continuous growth of load demands, the distribution systems operate closer to their limits. At the same time the liberalization of electricity markets drives them to decision making based more and more on electricity prices. Both aspects are influenced in a great extant by the value of technical losses in distribution networks. Some of the most efficient methods applied to reduce losses in distribution systems are: (a) balancing loads between the phases of the 3-phase system in the low voltage networks; (b) reconfiguring the distribution network in normal and post- outage conditions by changing its radial structure, and (c) controlling the reactive power to obtain desired voltage profiles and to change reactive power flow through the system. Reactive power control at the system level can be achieved using several approaches such as generator voltage control, transformer tap control and fixed or controllable VAR sources. At the distribution level, the most efficient approach, as it produces other positive effects too, is the Reactive Power Compensation (RPC) method through power factor correction. Current and power flows in distribution systems reduce through RPC. For instance, a reasonable 10% reduction in the current flow produces a 19% reduction in branch losses. Therefore, the RPC approach might be an important source for loss reduction and financial savings for any DISCO. In addition the RPC approach based on capacitor banks determines reduced voltage drops in the system and decreases voltage fluctuations. This paper considers the RPC problem as finding an appropriate placement of reactive power sources (capacitor banks) to minimize system losses, while ensuring predefined voltage profiles in the buses of the distribution system. Inequality constraints are also considered to account for minimum and maximum voltage levels or the maximum allowable power supplied by reactive sources. Traditional techniques use linear or/and non-linear programming or gradient descent optimization methods. The last two decades have shown an impressive development of new methods and algorithms based on stochastic methods and computational intelligence. These approaches have widely used simulated annealing, genetic programming or genetic algorithms [3,6]. Recently new computational intelligence approaches, based on immune algorithms [5] and particle swarm optimization techniques [2], were proposed. All these techniques aim to solve complex optimization problems by applying models and mechanisms inspired from natural selection, immune systems or swarm behavior. This paper presents a comparative study for the RPC problem using three optimization methods based on computational intelligence techniques, namely genetic algorithms (AGs), immune algorithms (IAs) and particle swarm optimization (PSO) algorithms. A new searching strategy is also proposed to enhance the convergence properties of the standard PSO (S-PSO) algorithm. PROBLEM FORMULATION The RPC problem through optimal placement of reactive sources aims to identify an optimal solution for the placement of a stock of capacitors in the nodes of a distribution system with a known configuration that minimize an objective function. The unknown variables are locations, type and ratings of capacitors. This problem is one of the most complex optimization problems in distribution systems as it acts on an essentially non-linear objective function (power losses or combination of power losses and other functions for multiple criteria optimization) and must comply with non-linear inequality constraints. The objective function addresses three optimization criteria: (i) minimizing system energy losses during a given period of time; (ii) minimizing voltage deviations with respect to the system rated voltage and (iii) minimizing the number of shunt capacitors installed in the system. On the other hand, three types of constraints must be controlled: (i) keeping the system voltage between minimum and maximum values; (ii) excess reactive power compensation in the nodes of the distribution system is forbidden and (iii) number of capacitors installed in the system is limited by an existing stock. The optimization criteria and the constraints are included in a unique objective function: C C U U U U W W obj P P F F F ⋅ + ⋅ + ⋅ + ⋅ = β β α α (1) where: F obj – the global optimization function of the problem; F W – partial objective function for the system