a b s t r a c tDistribution feeder reconfiguration has been an active field of research for many years. Some recent theoretical studies have highlighted the importance of smart reconfiguration for the operating conditions of such radial networks. In general, this problem has been tackled using a multi-objective formulation with simplified assumptions, in which the uncertainties related to network components have been neglected by both mathematical models and solution techniques. These simplifications guide searches to apparent optima that may not perform optimally under realistic conditions. To circumvent this problem, we propose a method capable of performing interval computations and consider seasonal variability in load demands to identify robust configurations, which are those that have the best performance in the worst case scenario. Our proposal, named the Interval Multi-objective Evolutionary Algorithm for Distribution Feeder Reconfiguration (IMOEA-DFR), uses interval analysis to perform configuration assessment by considering the uncertainties in the power demanded by customers. Simulations performed in three cases on a 70-busbar system demonstrated the effectiveness of the IMOEA-DFR, which obtained robust configurations that are capable to keep such system working under significant load variations. Moreover, our approach achieved stable configurations that remained feasible over long periods of time not requiring additional reconfigurations. Our results reinforce the need to include load uncertainties when analyzing DFR under realistic conditions.

Distribution network reconfiguration problem is simply aimed at finding the best set of radial configurations among a huge number of possibilities. Each solution of this set ensures optimal operation of the system without violating any prescribed constraint. To solve such problem, it is important to count on efficient procedure to enforce radiality. In this paper, we propose a reliable approach to deal properly with topology constraint enabling algorithm convergence toward optimal or quasi-optimal solutions. A simple and practical codification of individuals used in evolutionary algorithms to solve the generalized reconfiguration problem is detailed in this paper. Mathematical formulation, algorithm, and simulation results are presented for the distribution reconfiguration problem, incorporating a new representation scheme which is immune to topologically unfeasible possibilities. The individual interpretation procedure is straight and it demands no additional data structure or graph preprocessing. Comparisons are made for five well-known distribution systems to demonstrate the efficacy of the proposed methodology. It is also demonstrated that optimal configurations are properly surveyed when single or multiple sources are dealt.

No abstract