A novel and efficient method for increasing distribution systems reliability via network reconfiguration

Abstract

In our electricity-dependent era, power distribution systems are a vital infrastructures, that supports essential services. As demand grows, they face operational challenges, especially on contingencies. The constant expansion and irregular growth may lead to intricate, timeconsuming service restoration plans. Current operational topologies, designed for some level of efficiency under normal conditions, are likely ill-suited for contingencies. This paper proposes a novel grid reliability methodology, where the sum of SAIDI and SAIFI indexes over numerous fault scenarios provides a reliability measure. The optimization goal is to minimize these indices, to find an operational topology that excells in both normal efficiency and contingency reliability, that respects operational restrictions. Several performance enhancements are developed to make this computationally intensive task manageable, such as the use of Evolutionary Algorithms, Parallelization of Restoration simulations, Monte Carlo Method for significant fault cases, and to avoid fitness calculations on context-known topologies. Validation on small and large scale grids demonstrates the methodology’s efficiency: results show more reliable grids than the initial configuration, with significantly lower SAIDI index in tailored training fault cases; also, the new topology reflects better SAIDI indexes for other unknown evaluation fault cases. The performance enhancements proved to be scalable and necessary for timely computations of the reliability problem even in small grids scenarios with plenty computation.