K - Medoids Clustering of Setting Groups in Directional Overcurrent Relays for Distribution System Protection

Large-scale integration of Distributed Energy Resources (DERs) in power distribution networks has raised concerns for distribution system's protection coordination. Due to the bi-directional flow of power, Directional Overcurrent Relays (DOCRs) are being widely used in distribution systems. The pickup settings and time dial settings of these DOCRs needs to be modified with changing topologies and added DERs, to ensure proper selectivity and sensitivity towards fault identification and isolation. However, the setting groups that can be stored in DOCRs to accommodate these numerous topologies are limited. The work in this paper discusses the implementation of K-Medoids clustering technique to cluster the numerous grid-connected and islanded topologies of a distribution system based upon the various pickup currents and operating time indices. The testing results generated on a modified IEEE 13 node test feeder system demonstrate the effectiveness of the methodology over prior works.