DC fault detection and clearance in VSC-HVDC systems

• Main Author: Yeap, Yew Ming
• Other Authors: Wang Youyi
• Format: Theses and Dissertations
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electric power; DRNTU::Engineering::Electrical and electronic engineering::Power electronics
• Online Access: http://hdl.handle.net/10356/74388
• Institution: Nanyang Technological University

Today’s electric grid is mostly based on alternating current (AC), as well as the loads. The advancement of voltage source converter (VSC) has paved way for the emergence of VSC-HVDC transmission systems opening up the possibility to build multi-terminal DC (MTDC) grid in future. Before we reach that mark, there are a lot of open points to be answered. One of the critical topics for such DC grid is the protection. DC fault can be identified by the time- and frequency-domain analysis of the DC current. Thus, this thesis proposes and investigates three fault detection methods which operate in the two domains. The first frequency-domain method involves wavelet transform. Because DC fault can happen anytime and is unpredictable, the signal processing technique can work on the non-stationary fault signal and identify the presence of high frequency content caused by fault. The second method is based on time-domain, whereby the behaviour of DC-link capacitor in VSC is closely monitored. In the event of fault, the DC-link capacitor will discharge contributing to high fault current. Short-time Fourier transform (STFT) is another frequency-domain method to be considered. It will be studied how the distortion in frequency spectrum can be used to establish the indication of DC fault occurrence. Since wavelet transform has been well-established in literature, it will be used as the benchmark to assess the performance of capacitive discharge and STFT methods. They are extensively analyzed and compared using a scaled down VSC-based DC system experimental test setup, as well as simulation model built in PSCAD/EMTDC. Results show that capacitive discharge performs better in terms of detection time and robustness, whereas STFT method consumes less computational resource, yet it can detect the fault as effective as wavelet transform. It will be shown how the noise in real fault signal affects their performances to varying degrees. Besides that, the proposed methods succeed in determining the faulted line in MTDC systems by just relying on local information. It is also equally important to address the issue of fault clearance in DC systems. This thesis will discuss and analyse three protection schemes. Each scheme will be assessed based on how fast they clear the fault and achieve smooth recovery. For this study, it is entirely done in PSCAD/EMTDC.