Locating Method for Electrical Tree Degradation in XLPE Cable Insulation Based on Broadband Impedance Spectrum

Abstract

Electrical treeing is one of the main causes of crosslinked polyethylene (XLPE) cable failure. The current methods for locating electrical trees are mainly based on the partial discharge (PD) signal. However, PD signals are easily attenuated in the long cable and the PD test voltage may cause damage to the insulation. This work proposes an improved broadband impedance spectrum (BIS) method to locate electrical trees in XLPE cable. A mathematical model of a long cable containing local electrical tree degradation is established. The Gaussian signal is chosen as the simulated incident signal to reduce the spectral leakage. The location spectrum is obtained by multiplying the frequency domain function of the single-ended reflection coefficient and the Gaussian pulse. It has been found that the location spectrum of the local capacitance change can be characterized as a typical double-peak waveform and the spectrum of the local conductance change can be regarded as a typical single-peak waveform. Electrical tree experiments at different temperatures were carried out to initiate different types of electrical trees. A vector network analyzer (VNA) was used to test the high frequency capacitance characteristics in the treeing process. The location spectra of the 20 m long cable containing different types of electrical trees was calculated by the improved location algorithm. The results show that the location error of local electrical tree degradation is less than 3%. The capacitance of the sliced sample decreases with treeing time. The effect of the bush-pine tree on capacitance parameters is greater than that of the branch-pine tree. A typical double-peak is found in the bush-pine tree location spectrum and a single-peak is found in the branch-pine tree spectrum.