Modeling and detection of transformer internal incipient fault during impulse test

Abstract

Incipient faults are low magnitude and fast decaying transients, lasting only for a few microseconds during impulse tests of transformers. They are very difficult to detect, even for experts, and may cause severe problems leading to faults and consequent outage. Thus, simulation and diagnosis of incipient faults is very important requiring a valid model and advanced detection technique. In this paper, a new arc discharge model is proposed to simulate arcing phenomena between two discs of a power transformer winding, as one type of important incipient fault, during impulse test. Using wavelet transform techniques, a new methodology is developed. It is shown that winding input current, affected by arcing phenomena and obtained through impulse voltage test, provides important information for detection and analysis of this transient type of minor fault. A number of experiments and simulations were performed to evaluate the validity and accuracy of proposed model and diagnostic technique.