A study of PD location issues in GIS using FDTD simulation

An effective way of measuring partial discharge (PD) for GIS is the UHF method, by which both PD detection and location can be realized. PD location is based on time-off-light calculation in which travelling paths of electromagnetic signals between two sensors are assumed to be axially along the centre of GIS. The location is therefore normally obtained by solving a one dimensional problem, as if the GIS was in a continuous straight line. However, usually GIS must be made very compact, which can require many right angle bends or tee sections. The effect of corners and tee sections is to give greater significance to the three dimensional character of the PD location space. In particular, when a sensor is located close to a tee section, depending on where it is mounted on the tank circumference, the location result may vary because it does not consider the three dimensional relationship between this sensor and another sensor in a different arm of the tee section. This paper describes an investigation into the PD location accuracy in GIS tee sections using the Finite Difference Time Domain (FDTD) method to simulate the propagation of electromagnetic signals radiated by a PD. A tee section model of GIS was defined and a PD current pulse measured in SF6 was used to quantify the waveshape of the excitation source. Simulations were carried out and electric fields were monitored at various points where UHF sensors might be installed around the tee section. PD location was evaluated by time-of-flight calculation and the location errors were analyzed. By these means, potential inaccuracies of PD location in GIS were investigated and quantified. Conclusions were drawn regarding the positioning of UHF sensors to reduce the location error.