Fault location in resonant earthed medium voltage distribution systems
Time: Fri 2022-12-16 10.00
Location: D2, Lindstedtsvägen 5, Stockholm
Video link: https://kth-se.zoom.us/j/68432985723
Subject area: Electrical Engineering
Doctoral student: Md Zakaria Habib , Elektroteknisk teori och konstruktion, Swedish Energy Agency (SWEGRIDS)
Opponent: Professor Matti Lehtonen, Aalto University, Helsinki, Finland
Supervisor: Associate professor Nathaniel Taylor, Elektroteknisk teori och konstruktion
A major challenge in electric distribution systems is to increase the reliability of the electricity supply in an economical way. At the final medium-voltage level in the system, such as 10 kV or 20 kV, a single feeder may supply tens of medium/low-voltage transformers and hundreds of customers, with many branches of lines and cables where faults can occur. Resonant earthing is often used for these networks in Sweden, enabling earth-fault currents to be kept at low levels such as 10 A. This brings some advantages with safety and with avoidance of supply disruption during transient faults. However, the low fault currents, much less than typical load currents, make it hard to locate faults. In some countries, this type of system could be operated while an earth fault is traced, or a lower-impedance could be switched into the system earthing to increase the fault current to a clearer level. In Sweden, the legal safety requirements in most types of medium voltage networks require prompt disconnection of earth faults, which means that a single earth fault anywhere in an extensive medium voltage network can cause an outage for all customers on its feeder or section until it is located and repaired. Several approaches have been taken by network companies to reduce these outage times and thereby increase supply reliability.
The focus of this thesis is to develop methods for determining where a fault is, in order to speed up both the repair process and any switching operations that can restore supply to customers outside the faulty section. Such methods include true fault location (FL), which gives an estimated distance to a fault, and fault-passage indication (FPI), which tells whether a fault is detected downstream of a particular point. A comparison of FL and FPI is made, showing that classic central FL needs an accuracy beyond what is currently available, if it is to give a better improvement in supply reliability than a small number of well-placed FPI units. Three algorithms suitable for FPIs are proposed in this thesis that use only current measurements from different parts of the network to identify the faulty section. One of the methods is based on the information from the phase angles of the healthy phases relative to the faulty phase. It shows good results for under-compensated and high load conditions but struggles for feeders with high charging currents. The next method uses incremental phase currents and shows promising results for over-compensated cases but unreliable results for under-compensated cases. The third method uses the zero sequence current magnitude and the location information of the measurements. It shows good results for the homogeneous feeders in both over- or under-compensated cases. A method for estimating the fault distance using multiple measurements from the network is also presented and tested. Its main trouble is its sensitivity to the fault resistance.