Single-objective probabilistic optimal allocation of capacitors in unbalanced distribution systems

• Published in: Electric power systems research 2012-06, Vol.87, p.47-57
• Main Authors: Carpinelli, G., Noce, C., Proto, D., Russo, A., Varilone, P.
• Format: Article
• Language: English
• Subjects: Applied sciences; Electrical engineering. Electrical power engineering; Electrical power engineering; Exact sciences and technology; Linearization method; Micro-genetic algorithms; Miscellaneous; Optimal allocation of capacitors; Point estimate method; Power networks and lines; Probabilistic formulation
• ISSN: 0378-7796; 1873-2046
• DOI: 10.1016/j.epsr.2012.01.008

► We consider the optimal allocation of capacitors in unbalanced systems. ► We give a probabilistic formulation of the optimisation problem. ► The solution procedure is based on the application of micro-genetic algorithms. ► The linearisation and the point estimate methods are applied as well. ► We apply the solution procedure to the IEEE 34-node test system. The optimal allocation of capacitors in unbalanced distribution systems can be formulated as a mixed integer, non-linear, constrained optimisation problem. Fuzzy-based approaches, simulated annealing, tabu search and genetic algorithms are some of the techniques used for solving the problem in deterministic scenarios. However, distribution systems are probabilistic in nature, leading to inaccurate deterministic solutions. As a result, a probabilistic optimization model is required to take into account the unavoidable uncertainties affecting the problem input data, primarily the load demands. Of the various techniques for the solution of the problem, one of the most frequently used is the genetic algorithm. However, the application of simple genetic algorithms to solve the probabilistic optimization model involves tremendous computational effort. To reduce the computational effort, this paper proposes a new single-objective probabilistic approach based on the use of a micro-genetic algorithm. Two different techniques, one based on the linearised form of the equality constraints of the probabilistic optimisation model and one based on the point estimate method, were tested and compared. The proposed approaches were tested on the IEEE 34-node unbalanced distribution system to demonstrate the effectiveness of the procedures in generating reduced computational efforts and increased accuracy of the results.