Comparing the New Improved RLC and CMTL Models for Measuring Partial Discharge in Transformer Winding

Nowadays, one of the most well-known methods for accurate simulation of the fast-transient phenomenon as well as the study and location of partial discharge (PD) in transformer windings is the method based on winding modeling. In this regard, the circular multi-conductor transmission line (CMTL) model, which is an improved model of the MTL model, has been proposed as the finest solution since it is acceptable in an extensive frequency range and it can investigate the effect of winding rotation by considering the radius and angle variables. However, one of the fundamental problems of using the CMTL model is that a set of large equations, known as complete admittance matrices, must be solved for all winding turns, which typically contain thousands of turns. To overcome this problem, we presented an improved RLC model, which has the same accuracy as the CMTL model in terms of frequency validity range and less computational complexity. In this model, the variables of length and curvature of the turns are also considered in the formulations. Although this method is theoretically accurate, its reliability method must also be determined in practice and how the PD signal propagation must be carefully considered due to the problems in high-frequency modeling of the winding. Numerical analyses and laboratory measurements resulting from the use of PD pulses on a 20 kV winding of a distribution transformer, as well as a 63 kV, 30 MV winding, revealed that a valid frequency range is extendable up to 5 MHz for the proposed improved model without a significant increase in computational complexity.

Automated summary

The article introduces the method of accurate simulation of fast transient phenomena and partial discharge research based on winding modeling, proposing the CMTL model and improved RLC model as solutions. The improved RLC model is shown to have the same accuracy as the CMTL model in terms of frequency validity range and computational complexity.