On the Computation of Frequency-Dependent Core and Proximity Effects for Transient Analysis of Transformer Windings

Overvoltages and dielectric stresses caused by transient conditions can result in the failure of transformer windings. Adequate prediction of these conditions is generally based on the implementation of accurate computer models, which, in turn, require an accurate computation of electrical parameters. This paper analyzes the computation of frequency-dependent core and proximity losses included in the transient model of transformer windings, using the finite element method and analytical expressions. A modified formula to calculate the proximity losses is presented. The results from a frequency-domain winding model are compared with experimental measurements on a disk-type prototype winding. These results show that including the frequency-dependent proximity losses is essential to accurately reproducing the experimental results. They also show that the core losses have a modest effect on the transient response of the winding when compared with measurements.

I. Introduction

According to CIGRE's classification of electromagnetic transient models [1], fast transients are characterized by fast front pulses and frequencies up to 3 MHz, while very fast transients exhibit frequencies up to 50 MHz. Modern transformers are subjected not only to fast front excitations due to lightning and switching conditions, but also to fast and repetitive pulses related to the widespread inclusion of power electronic components [2]–[4]. These phenomena can produce large transient overvoltages and dielectric stresses that can damage the transformer or significantly reduce its life expectancy [5]. The definition of models able to accurately and effectively predict these phenomena is essential for the insulation design of transformers. Taking into account that, in general, the operating lifetime of a transformer is directly related to the aging/deterioration process of its insulation system, any accuracy improvement of transient models can have a great impact on the design of efficient and reliable transformers.