Protection location in distribution networks using particle swarm optimization

Abstract

Circuit breakers in radial topology distribution networks protect the system against short circuits. In addition, other switching devices are usually used to reconfigure the network in post-fault scenarios, thus reducing the number of affected users. This strategy is intended to reduce the loss of profits for the distribution company, as well as the economic penalties it must face due to the energy not supplied (ENS) under these conditions. This work proposes the implementation of a particle swarm optimization algorithm (PSO) to determine the best location of new protection circuit breakers to be incorporated in the system, in order to minimize the total annual ENS, without violating usual operational constraints. For this purpose, it is combined with other procedures designed to simulate faults on a mathematical model of the network, determine the operating protection and the best possible reconfiguration. The applications are tested on a network in the province of Santa Fe (Argentina), from which it is shown that an inadequate location of the new protections may result in counterproductive situations in terms of the obtained ENS. On the other hand, the tests show that the developed methodology converges to the optimal solution with substantially lower computation times than those demanded by other evaluated alternatives, such as direct search solutions and Monte Carlo simulations. To improve the performance of the PSO, it is combined with a clustering technique that divides the solution space into smaller subdomains, thus enabling a more effective exploration by the algorithm.