Overhead line partial discharges detection

Summary

On the one hand, high voltage assets are very expensive and one could not consider a replacement policy only based upon their age. On the other hand, power networks are crucial to our modern societies and we do want them to be extremely reliable. In this context, condition monitoring has been considered as a powerful tool of asset management and his development is of great interest. One of the most interesting condition monitoring techniques is partial discharges (PD) detection and analysis. This technique is well known and developed for a lot of high voltage components. However, there is still very few scientific work or practical applications for overhead lines. This paper presents a brand-new way to detect and measure PD directly on an overhead line bare conductor. At first, the partial discharges sensor has to be selected. After having performed a deep PD analysis, a very specific and appropriated current transformer is chosen. It presents a bandwidth and a sensitivity that suit pretty well the PD characteristics. Also, the PD measurement unit has to be built. The challenging detection of very small and high frequency signals has been made possible thanks to the design of a shielded measurement unit referenced at the high potential of the conductor. The unit is completely isolated from the ground and can be remotely controlled. Then, experiments have been carried out in a high voltage laboratory and PD patterns have been displayed. Both healthy and faulty conductors have been considered and they could be differentiated thanks to the PD patterns. Finally, another experimental measurements have been performed on a real overhead line 22 kV to investigate propagation and attenuation of PD under noisy condition of distribution power system. Very short impulses were generated and transmitted to the overhead line when voltage and current waveforms were recorded at the beginning of line and at the end. Final attenuation characteristics are determined after data signal processing. Those results are encouraging and can lead to an efficient and online overhead-line monitoring technique. The steps towards such a tool are under progress and are conveniently discussed.