Advances in Electrical and Computer Engineering Volume 11, Number 2, 2011 A New Method for Detection and Evaluation of Winding Mechanical Faults in Transformer through Transfer Function Measurements Mehdi BIGDELI 1* , Mehdi VAKILIAN 2 , and Ebrahim RAHIMPOUR 3 1 Department of Electrical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran 2 Department of Electrical Engineering, Sharif University of Technology, Tehran, Iran 3 ABB AG, Power Products Division, Transformers, Bad Honnef, Germany * Email: m.bigdeli@srbiau.ac.ir 1 Abstract—Transfer function (TF) is an acknowledged method for power transformer mechanical faults detection. However the past published works mostly discovered how to specify the faults levels and paid less attention to detection of the type of faults using comparison of TFs. whereas, it seems important for most of the applications to specify the type of fault without opening the unit. This paper presents a new method based on vector fitting (VF) to compare the TFs and specify the type, level and location of the fault. For development of the method, and its verification the required measurements are carried out on four model transformers; under intact condition, and under different fault conditions (axial displacement, radial deformation, disc space variation and short circuit of winding) and the TFs are determined. Employing VF, the coefficients of TFs are determined with the required accuracy. Using those coefficients, a new index is introduced to specify the type, level and location of the fault in the winding. Convincingly good results were obtained. Therefore it is believed that this finding could be helpful in fault diagnosis in actual power transformer windings. Index Terms—Transformer, Fault Diagnosis, Measurement, Transfer Function, Vector Fitting I. INTRODUCTION Power transformers are among the most important equipments in the electrical power transmission and distribution systems and occurrence of any fault in these transformers will reduces the power system reliability, in other words increases the possibility of power supply interruption. In addition, it will impose extra costs required for their maintenance and for their transportation to a repair factory. Due to the existence of a strong competition in the electrical power supply industry, the importance of application of the transformer monitoring systems has increased over the time [1]. Different fault modes in transformers are classified into two main groups; the mechanical types, and the electrical types. Different methods have been offered for detection of each of these faults, such as: a – Oil analysis (Dissolved Gas Analysis, Furfural) b – Partial discharge measurements c – Dielectric response analysis d – Transfer function analysis. Among these methods, TF method is a well-known method for detection of the winding mechanical faults [2]- [4]. TF method is a comparative method; in this method a new measurement is put besides a referential measurement. If the deviations were remarkable, the direction and magnitude of deviations should be studied and analyzed. Mechanical faults such as; axial displacement (AD), radial deformation (RD), disc space variation (DSV) and short circuit (SC) of the winding affect the TFs and cause dislocation of resonance frequencies and moreover will decrease or increase the magnitude of resonance frequencies in the TF. These changes will cause a relevant shift in the poles and zeros of TF. Therefore, by evaluation and comparison of the TFs, any variation in the structure of the winding due to a mechanical damage can be detected. TF evaluation and its comparison can be done in time domain or in frequency domain. The identification process is normally carried out in frequency domain through inspection of the rational functions. Some studies [5]-[6] have shown that VF method is an effective and accurate method for estimating the transformer TFs. In [7]-[8] have proposed an index (based on TF estimation using VF) for determination of AD and RD extent. That index played its role only to specify the level of AD or RD and does not specify the type of fault. In addition, their proposed index only perceives the variations in the poles of the rational function where as, during incidence of winding mechanical faults both the poles and the zeros change. Therefore continuation of the researches on VF application for fault detection is necessary. In this paper employing VF, the TFs of the transformer are estimated both during faulty condition and during normal condition. Introducing a new algorithm the TF is identified and compared with the reference one. Not only this algorithm is able to recognize the type of fault, but also it can specify the level and location of fault. Comparing this new method with one of the past methods, it is shown that the proposed index is more efficient, even in determination of the fault level. II. LITERATURE REVIEW AND PROBLEM DEFINITION TF analysis is a reliable method for winding failures detection in power transformers. Different methods are introduced for evaluation of the TFs. Some of these methods are verified against the practical samples of mechanical 23 [Downloaded from www.aece.ro on Wednesday, June 01, 2011 at 16:04:09 (UTC) by 217.218.226.142. Redistribution subject to AECE license or copyright. Online distribution is expressly prohibited.]