Lightning Protection with a Differentiated Arrester Configuration for Distribution Networks Using a Multi-objective Optimization Procedure

In addition to insulator flashover, lightning damage of surge arresters (SAs) is also considered in evaluating the lightning performance of a distribution network. This paper presents a novel multi-objective optimization framework for designing differentiated SA protection, by minimizing both the flashover number and SA damage rate. An artificial neural network and an energy-distribution surrogate model are developed for the efficient evaluation of these two targets in an optimization process. The samples for building these models are generated with a hybrid simulation approach together with a Monte Carlo method. It is concluded that the surge energy into an SA generally follows a log-normal probabilistic distribution. The mean of energy absorption decays exponentially with the distance to the stroke location. Differentiated SA configurations for a 10kV distribution network are then identified by balancing the need of minimizing these two targets. It is found that when increasing the SA number, the damage rate becomes dominant and SAs are recommended to be located close to the dead-end poles in sub-feeders. Otherwise, the annual flashover number becomes dominant and SAs tend to be located around the junction poles on the main feeder. When the annual flashover number is primarily addressed, the pole-group selection strategy is the best one due to its simplicity and better performance.