Protection and DC Fault Location for MT-HVDC Transmission Systems

Abstract

This thesis explores a Travelling Wave (TW) parameter fitting-based approach for both fault location and protection in a multi-terminal High Voltage Direct Current (HVDC) network. Utilizing a three terminal HVDC network model within a Real-Time Digital Simulator (RTDS) environment, a custom parameter fitting control component is developed using the Adaptive Multi-step Levenberg-Marquardt (AMLM) algorithm. For fault location, the methodology analyzes the line-mode backwards travelling voltage wave (Vb1) following an internal DC cable fault. Despite the successful integration of the AMLM parameter fitting algorithm into the real time environment, the achieved average absolute error of 9.80% (29.4 km) for faults spaced 50 km along a 300 km cable was above the acceptable threshold, rendering the proposed
method impractical for precise fault location. Improvements through specific signal truncation and optimal cable parameter selection reduced the error to 5.15% (15.45 km), which remains insufficient for practical applications. In contrast, the TW parameter fitting method proves highly effective for fast and fully selective protection. The protection scheme accurately discriminates between internal and external faults using the Vb1 signal and determines fault types through Vb0 signal analysis. Extensive testing revealed a fault detection rate of 100%, with an overall accuracy of 99.91% for fault resistances up to 200Ω. Severe internal faults are isolated in 1.68 milliseconds, while non-severe internal faults are typically isolated in 3.84 milliseconds. The method completely eliminates relay deadzone, providing robust performance even under noisy conditions with an accuracy of 99.85% for faults with an impedance up to 50Ω. These findings highlight the potential of the TW parameter fitting approach to significantly enhance the reliability and promptness of fault isolation in HVDC systems, while offering insights into the challenges and limitations of fault location accuracy on a real time platform.