Fault detection in a power transformer based on reverberation time

Abstract

The power transformer is one of the vital elements of electric power systems. The reliability of power transformer operation is directly related to its condition. Detection of winding deformations is possible by means of sweep frequency response analysis. Interpretation of SFRA results is usually performed by visualization of images generated using specialized devices and comparison of transfer functions obtained from successive tests. This diagnostic method is known as a fingerprint and its disadvantage is the subjectivity and relying on engineer’s experience. This paper presents an experimental approach for detection of internal transformer short-circuit faults. Existing mathematical indicators based on frequency response analysis are often unable to unambiguously indicate the degree of fault and their determination requires high-cost measurement and signal processing equipment. A new indicator based on the reverberation time, commonly employed in acoustics, is proposed instead. The proposed parameter is based on time-domain calculation and enables reliable detection of transformer inter-turn short circuits and accurate estimation of their degree. A significant advantage of the proposed approach is its low dependence on the frequency range of the measured signal, which allows employment of low-cost measuring and signal processing devices.