Fast and Reliable High-Impedance Fault Detection in AC Distribution Systems
Abstract
Detecting High-Impedance Faults (HIFs) are challenging due to their low short-circuit current magnitude. An undetected HIF may jeopardize human safety and cause arcing and fire ignition. Therefore, HIFs must be detected and cleared in a timely manner to ensure human safety and to avoid fire hazards. The conventional relays and protection systems often fail to reliably detect and clear HIFs. Moreover, the existing HIF detection methods: (i) often mis-detect non-HIF disturbances as HIF, (ii) typically fail to detect HIFs in a timely manner, (iii) typically are unable to detect HIFs distant from the relay location (Remote HIFs).
This thesis focuses on developing fast and reliable methods for detecting High-Impedance Arcing Faults, hereafter referred to as High-Impedance Faults, in medium-voltage AC distribution systems. The proposed HIF detection methods, which are based on digital signal processing and machine learning techniques, are evaluated using various case studies including (i) simulated HIFs on seven different fault surfaces, (ii) simulated non-HIF scenarios such as capacitor switching, operation of nonlinear loads, and transformer energization, (iii) recorded HIF signals obtained from experimental HIF tests on four different fault surfaces. The time-domain simulation studies are conducted on the IEEE 13-node test feeder using the PSCAD/EMTDC software. The study results indicate that the proposed HIF detection methods (i) can detect all investigated HIF scenarios under 200 ms, which is well-below the typical HIF detection speed reported in the literature, (ii) are sensitive to remote HIFs, (iii) do not cause false tripping against non-HIF scenarios (iv) are computationally efficient.