Simple

Fault Detection in Mesh Network Utilizing Measured Data from MV/LV Transformer Stations

*Vit Krcal
Dept. of Electrical Power Engineering Faculty of Electrical Engineering and Communication, Brno University of Technology

David Topolanek
Dept. of Electrical Power Engineering Faculty of Electrical Engineering and Communication, Brno University of Technology

Abstract
Nowadays, the integration of smart grid components in distribution networks is thriving and various measurement units are being deployed in distribution systems, even on the low voltage level. Data measured on such devices should be used for enhanced monitoring functionalities to provide better insight of the network status to the DSO and enable e.g., to optimize the operation. These functionalities include detection and localization of faults and abnormal events, which is beneficial especially in mesh networks, which are very complex and is difficult to estimate stat od their components.
In this paper, extensive MV/LV mesh network numerical model, which consists of 1268 nodes, is considered. Based on previous research and current trends of the DSOs, the measuring units placement is emulated at MV/LV transformer stations, where electrical currents are measured in all MV/LV stations and in all LV feeders. A simple deterministic algorithm is designed to detect abnormal states in the LV network. The detection principle is based on evaluation and comparison of changes of current in LV feeders and total currents in transformers. The study is complemented with a sensitivity analysis of impacts of loads behaviour to monitored current changes, which is important for differentiating between normal and abnormal states.
Performance of designed simple algorithm is tested on numerical model of extensive distribution mesh network, where numerous fault scenarios are simulated as well as load stochasticity for several network configurations. The paper presents an option how to simply assess abnormalities occurrences in large mesh network with measuring currents in all MV/LV stations. The results are recommendation for correct setting and enhancing the algorithm as well as findings of sensitivity analysis of detection to load changes.