Internal Fault Analysis in Disk-Type Transformer Winding Using Network Reduction

Abstract

Internal winding faults are the most common cause of failure in power transformers. The complicated electrical and geometrical configuration of transformers makes it difficult to produce models able to predict their behavior under transient conditions in an accurate and efficient manner. Internal fault analysis of windings requires a detailed (turn-by-turn) modeling approach that results in very large systems that can be difficult to define and computationally expensive. In addition, in many cases the fault analysis does not require from measurements at all turns, but only at the turns related to the fault condition. The purpose of this work is to describe a modeling approach for internal fault analysis of transformer windings using a frequency domain approach in combination with Kron's network reduction to preserve only the nodes of particular interest, and the numerical inverse Laplace transform to obtain the transient response of the system in the time domain. A prototype transformer winding with 2000 turns is used for this purpose. Comparisons between the open and short circuit faults for the complete network system and the equivalent reduced network provide the same results considering a very small portion of the original system's size.