1 A Novel Inductance-Based Technique for Discrimination of Internal Faults from Magnetizing Inrush Currents in Power Transformers H. Abniki, H. Monsef, P. Khajavi, and H. Dashti Faculty of Electrical and Computer Engineering University of Tehran Tehran, Iran hnabniki@ut.ac.ir Abstract— This paper presents a new methodology for discrimination between inrush currents and internal faults for a three-phase power transformer. This algorithm is based on instantaneous inductance. First, this method calculates differential inductance of transformer phases from primary side view of transformer by using voltage and current signals. Then, the algorithm compares differential inductance with a threshold value. If the calculated criterion is over than the threshold, disturbance will be inrush current. Otherwise, if the calculated criterion is lower than the threshold, disturbance will be internal fault. The operating time is less than 5 ms, less than 1/4 rd of power frequency cycle. The scheme uses voltage and current signal; hence it can work reliably in the presence of transformer tap variation, fault resistance and as shown later, the criterion works properly in CT saturation or over-flux condition. Simulation studies with respect to different faults and inrush conditions have been conducted and the results proved that the proposed technique is able to offer fast responses in protection and accurately discriminate between inrush currents and internal faults. Keywords- Equivalent inductance, inrush current, internal fault, transformer protection. I. INTRODUCTION It is obvious that performance of power systems directly relates to the power transformers. In this regard, transformer protection shows itself essential case. The important objective of transformer protection is to detect internal faults from magnetizing inrush currents with a high and also good performance in all condition such as in external fault, in over- flux or CT saturation condition. There are different kinds of inrush current such as sympathetic inrush current or switching inrush current. Nevertheless, in some cases sensitive detection in transformer de-energization enables the faults damage to system and hence it is necessary to be limited. However, it should be able to provide backup protection through faults on the system, as these could lead to deterioration and accelerated aging or failure of the transformer. Winding insulation due to over-heating and high impact forces causes in the windings due to high fault currents. Moreover, abnormal system conditions such as over-flux, over voltage and loss of cooling about internal faults, can lead to deterioration of the transformer. Hence, protection again these failures should be considered as part of the transformer protection scheme. With the advent of electrical system, transformer protections are studied under various situations and conditions. Therefore, their protection can be vastly categorized not only as electrical protection implemented by sensing the current through it, but also voltage and frequency, as mechanical protection implemented by sensing operational parameters like oil pressure, gas evolved and winding temperature. In finance affairs, as every electrical engineer knows, in transformer protection too, the extent of protective devices applied to a particular transformer is dictated by the economics of the protection scheme, and the probability of a particular type of failure and the cost of replacing and repairing the transformer as well the possibility of the failure causing to adjacent equipment damage. Failure costs include all of the direct and indirect costs related to it. Disconnecting device, circuit breaker and other auxiliaries like batteries are necessary infrastructure and costlier protection schemes. Recently, many studies have done in transformer differential protection. An inductance technique for discrimination between inrush currents and internal faults is suggested in [1]. A morphological scheme for inrush identification in transformer protection was presented in [2]. Ref [3] proposed a new method to identify inrush current based on error estimation. In [4] a wavelet based method and in [5, 6], artificial neural network (ANN) was trained to distinguish between inrush and internal fault currents. In [7], a set of developed fuzzy laws based on differential current harmonics has been proposed as differential protection algorithm. Ref [8] proposed a sequential phase energization method for transformer inrush current reduction transient performance and practical considerations. One of the other methods operates via measurement of intervals between two successive peaks of differential current waveform is introduced in [9, 10]. The operation criterion in another method is the duration in which differential current waveform remains near zero. In some of these methods, differential current harmonics are used as inputs to a learned neural network [11]. The output of the neural network indicates transformer situation. Modern Electric Power Systems 2010, Wroclaw, Poland MEPS'10 - paper P32