Detection and localization of internal turn-to-turn short circuits in transformer windings by means of negative sequence analysis

Transcription

1 No. E-14-AAA-0000 Detection and localization of internal turn-to-turn short circuits in transformer windings by means of negative sequence analysis Malihe Abi, Mohammad Mirzaie Faculty of Electrical and Computer Engineering BabolNoshirvani University of Technology Babol, Iran Abstract Internal turn-to-turn short circuits are the most difficult types of faults to detect within the transformers. This paper proposes a new, simple and efficient algorithm in order to protect transformers against these faults. Using this protection algorithm which is based on negative sequence analysis, make it possible to detect and also locate low level inter-turn short circuits which typically cannot be detected by the traditional differential algorithm before they extended into more severe faults. Furthermore, the proposed algorithm is stable in the case of external faults as well as load imbalance condition. In this work, first, a typical transformer is modeled based on Finite Element Method to simulate the transformer behavior under different operation conditions. Then, the accuracy and performance of new protection technique in detection and localization of inter-turn faults is studied by applying it to the simulated transformer. Keywords transformer; turn-to-turn fault; Finite Element Method; negative sequence analysis I. INTRODUCTION Transformers are one of the most important and expensive devices in electrical systems that are critical links between the generation stations and consumers. Variety of unusual conditions and faults can affect the transformers. Unplanned repairs such as fix or replacement of the faulty transformer are very costly and time consuming. One of the most sensitive parts of the transformer is the insulation system that can be exposed by electrical, mechanical and thermal stresses and moisture. Degradation of insulation system causes a breakdown in the insulation and leads to development the inter-turn short circuits. Internal turn-to-turn faults are the most difficult types of faults to detect within the transformers. If turn-to-turn fault has not been rapidly detected, this fault can develop into more critical and costly to repair faults such as phase to phase or phase to ground faults. Therefore, quick detection of turn-to-turn faults is essential in order to protect the entire of electrical system and reduce the damage and repair cost. In this way, development of online techniques for condition monitoring and in order to diagnose of inter-turn short circuits is very important to improve the system reliability. Although the percentage differential relay is the most commonly used protection, it is not sensitive enough to detect low level turn-to-turn faults. Up to now, the advanced methods with high sensitivity such as Wavelet Transforms [1,], S Transform [3,4] and Hilbert Transform have been applied for turn-to-turn fault detection in transformer windings. Although these approaches detect the minute faults, but they are based on the signal processing and have complicated computation. Hence, these methods require large number of processors and instruments and so implementation of them is difficult [5]. In recent years, new diagnostic techniques based on sequence components have been attended because of their simplicity and also having a good sensitivity in winding faults detection. In 1998, sidhu and et al presented the earliest works in this field that uses the arguments of the positive and negative sequence impedances of the power system in a fault detection algorithm [6, 7]. In [8], authors differentiate between different transient states using1 th, th and 5 th harmonic components of the positive sequence differential current as inputs of artificial neural network. But, the methods based on neural networks require a large number of training patterns which are produced by simulation of various cases and this method is not generalized to be applied to different power transformers. Ref. [9] presented a digital technique for power transformer fault detection on the basis of positive sequence admittance approach. The proposed method employs the accumulated positive sequence admittances on the both side of power transformer, so that the contour of the accumulated values of positive sequence admittances computed by the relay is used for discrimination between internal and external faults. In [10, 11], a new method based on the phase difference between negative sequence currents at the two sides of power transformer was presented for diagnosing internal faults. But, the consequences of these works were also not convincing because of restricted studies carried out for only one set of system parameters and no detailed investigations. In [1], MariyaBabiy performed some simulations for various operating conditions and different configuration of the power transformer to study the performance of mentioned method. Ref. [13] offered a protection scheme for internal fault detection in power transformer using comparison the ratio of

2 negative sequence components on the primary and secondary side with the turn ratio of the transformer. In this paper, the fault detector was defined for only one type of transformer connection and proposed method is not generalized for other connections of transformers. It should be noticed that entire of the above mentioned schemes focuses on fault detection only and fault localization ability existing in traditional differential relay has been missed in these methods. To overcome the drawbacks of the existing fault diagnostic techniques, this paper focuses on developing a new and modified protection scheme using negative sequence components for detecting and also locating internal turn-toturn faults in transformer windings. To this end, a three phase transformer under normal operation as well as inter-turn fault condition has been simulated firstly based on Finite Element Method (FEM) in Maxwell 14.0 software. Then, the accuracy of suggested algorithm has been examined by applying it to the simulated transformer in various conditions and obtained results have been shown finally. condition has been applied to the boundary surrounding the transformer. In order to modeling the nonlinear nature of the transformer core, the magnetization curve of core material has been assigned to it. Afterwards, the circuit domain related to geometric domain has been designed in Ansoft Maxwell Circuit Editor environment. In the circuit domain, each separated region of coils in geometric domain has been modeled by its equivalent inductance calculated by software and the resistance corresponding to it. II. FINITE ELEMENT MODEL OF THE TRANSFORMER In order to simulate a turn-to-turn fault on the transformer winding, the defective transformer is modeled as shown in Fig. 1,. As can be seen from the figure, when a turn-to-turn short-circuit occurs on the transformer, the affected winding is divided in two subwindings a and b, that are associated to the healthy and faulty part, respectively. In this figure, the fault has been illustrated on the primary winding of phase A by connecting the fault resistance (RR ssh ) across the shorted turns. The severity of fault depends on the number of shorted turns as well as the circulating current flowing of them that is limited by the fault resistance. Fig.. Finite Element Modelof the transformer in Maxwell environment In order to simulate the inter-turn fault, the geometric domain should be modified as well as circuit domain. In the equivalent circuit a time-controlled switch has been used in series with the fault resistance to create the winding fault in desired moment. Finally, the simulated transformer has been analyzed by selecting the Transient mode for solution type and µs for time step (i.e. 10 KHz sampling frequency). The rated values of simulated transformer have been presented in TABLEI,. TABLE I. THE RATED VALUES OF THE TRANSFORMER Rated voltage ratio 0/0.4 KV Fig. 1. Equivalent circuit for turn-to-turn fault in transformer winding To investigate the transformer behavior in different situations, a two winding three phase transformer has been simulated in Maxwell 14.0 software which is based on Finite Element Method. The Finite Element Method is a quick and efficient way in simulation of modern engineering systems that solves a problem by separating the problem field into several elements and then implementing physical rules to each minor element. Fig., shows the Finite Element Model of aforesaid transformer in geometric domain of Maxwell software. Under healthy condition, each one of the transformer coils has been drawn as two whole regions and the nominal ampere-turn has been assigned to each region. For faulty transformer, the defective part of winding has been modeled as separated geometric region and desired value of ampereturn has been allotted to it. Then, the Dirichlet boundary Rated apparent power Rated frequency Connection style KVA 50 HZ Yzn5 Turn ratio 944/68 Core steel type III. PROTECTION SCHEME The proposed protection scheme for internal turn-to-turn faults is based on the negative sequence analysis. It is well known that only positive sequence component is always exist in normal operation but under faulty condition, negative and if possible zero sequence components are also create along with M5

3 positive sequence component and make clear indication of fault occurrence [10]. Use of negative sequence component has two significant advantages in comparison with zero sequence component that they are as below: Zero sequence currents will be eliminated by removing the ground from the unit. Negative sequence currents can be produced even when the fault does not include earth. In other words they can provide coverage for phase to phase and turnto-turn faults as well as ground faults [1]. When a short circuit including a few turns take place on the winding, the changes in magnitudes of the negative sequence currents on the both sides of the transformer compared to the steady state value of the phase currents are greater than the change in magnitude of the phase currents. Thus, negative sequence-based protection is a good complement to the conventional differential protection for detecting low level faults, with a high sensitivity and speed. A. Fault detection algorithm According to the principle of ampere-turns balance, sum of the ampere-turns on each one of the transformer core limbs is equal to zero at all times under any operating conditions. So, the following equation is always satisfied for a three phase transformer with Yzn5 connection (notation as per Fig. 1, ): ii LL1 NN 1 ii LL NN 1 ii LL3 ii eeeeee 1 ii eeeeee ii LL4 + NN ii ii LL5 = 0 eeeeee ii LL6 Whereii eeeeee 1,ii eeeeee,ii eeeeee 3 the three phase exciting currents, NN 1, NN turn numbers associated to the primary and secondary side of the transformer. The above equation is equivalent to: ii eeeeee 1 ii LL1 ii eeeeee = ii LL NN ii LL4 ii LL6 ii LL5 ii LL4 () ii eeeeee 3 ii NN 1 LL3 ii LL6 ii LL5 We can write aforementioned phrase in the phase domain as below: II eeeeee 1 II LL1 II eeeeee = II LL NN II LL4 II LL6 II LL5 II LL4 (3) II eeeeee 3 II NN 1 LL3 II LL6 II LL5 Symmetrical components corresponding to the primary, secondary and exciting currents are obtained by multiplying inverse of the Fortescue transformation matrix to the both side of (3), : II ee_zz II pp_zz II ee_pp = II pp_pp NN II ss_zz II ss_pp (4) NN II ee_nn II 1 pp_nn II ss_nn In the above equation, indexes e, p and s related to the exciting, primary and secondary currents and indexes z, p and n pointed to the zero, positive and negative sequence (1) components, respectively. Therefore, the principle of ampereturns balance is also valid for each one of sequence networks. According to (4), we have always the following phrase at the negative sequence network of the transformer: II ee_nn = II pp_nn NN II (5) NN ss_nn 1 The abovementioned phrase enumerates as important equation in negative sequence-based protection. Using the principle of ampere-turns balance can be achieved to the similar phrases for three phase transformers with various connections. In the proposed method, a fault detector index can be defined as the ratio between the negative sequence components related to the exciting and primary currents. It should be mentioned that calculation of the fault detector index has no need to calculate the II ee_nn separately and only measurement of II pp_nn and II ss_nn is enough. Dividing the both side of (5), to the II pp_nn yields (6), : II ee_nn II pp_nn = 1 NN NN 1 II ss_nn II pp_nn In this way, the fault detector index can be expressed as below: DDDDDDDDDDDDDDDD = 1 NN NN 1 II ss_nn II pp_nn % Logic of the proposed approach is shown in Fig.3,.The logical steps of proposed algorithm can be explained as below: All individual instantaneous currents on the primary and secondary side of the transformer have to be measured. Using the Discrete Fourier Transform (DFT) block determines the fundamental harmonic magnitude and phase of the input signal as a function of time. Calculate negative sequence currents on the primary side (I p_n ) and on the secondary side (I s_n ) of the transformer (i.e. magnitude and phase components of the fundamental harmonic) using Fortescue Transform. Check the magnitudes of negative sequence currents from both sides of the transformer and compare them with a pre-set level. The magnitudes of negative sequence currents from both sides of transformer have to be above the pre-set limit in order to calculate the fault detector index. The minimum pre-set level must be more than magnitudes of negative sequence currents that exist during the normal operation due to the prefault asymmetries of the power system. For the system studied in this paper, the minimum allowable negative sequence current (I min ) is 1% of the transformer rated current (i.e pu ). If the fault detector index exceed of the threshold (Thd) then an internal fault is happened and a trip command issued to prevent the fault extension. In this protection (6) (7) 3

4 algorithm, the recommended value for detector threshold has been considered 1%. Fig. 4. Negative sequence current phasor related to the primary side of the transformer in the case of turn-to-turn fault involving of turns in various phases of the secondary winding Fig. 3. Negative sequence current based logic B. Fault localization After doing the fault detection process by protection system, localization of fault in the transformer windings is one of the most important problems that solving it can be aid to perform preventive maintenance. Using a fault localization method, the faulty phase of the transformer is specified and consequently only one phase (i.e. faulty phase) needs to inspect and repair. In conventional differential protection, existence of non-zero differential current in a phase of the transformer points to the faulty phase, but the negative sequence based protections convert three phase variables to one variable (i.e. negative sequence current) and consequently, the fault localization in this methods will not be as simple as one that is performed in differential relay. In the proposed method, the negative sequence current phase on the primary side of the transformer after the fault occurrence can be used to determine the faulty phase. Simulation results show that when a turn-to-turn short circuit happens on each one of the transformer windings, I p_n changes its position in the coordinates plate and corresponding to its new position the fault location specifies in the transformer. Fig. 4, shows the negative sequence current phasor related to the primary side in the case of turn-to-turn fault involving of turns located on various phases of the secondary winding. It is observed that I p_n locates at the three separated regions of the coordinates plate that have phase difference near to 10 together and each region points to the fault occurrence in one of the three phases of the transformer. According to this observation, it seems that a fault localization stage can be added to the detection algorithm for determining the faulty phase of the transformer. The obtained results related to apply the proposed scheme for detection and localization of inter-turn faults in the simulated transformer will be presented in the next section. IV. SIMULATION RESULTS In this section, some test results verifying the performance of the protection scheme for various disturbance conditions such as internal turn-to-turn faults, external fault and also load imbalance have been presented. In all of the case studies, the disturbance occurs at t = ms. A. Internal turn-to-turn faults A large number of simulations were performed to prove the ability of proposed technique in detection of low level inter-turn faults, but only some chosen cases are contained here. It should be mentioned that R f has been adjusted to 1 µω for all of the inter-turn fault cases. 1.Turn-to-turn faults on the primary winding: Fig. 5, shows the negative sequence currents waveforms on the both sides of the transformer as well as fault detector index for a turn-to-turn fault involving of turns on the phase C of the primary winding. As expected, when the fault occur both negative sequence currents exceed of pre-set level and fault detector rise to %, therefore a fault alarm is activated.. Turn-to-turn faults on the secondary winding: When an inter-turn fault takes place on the secondary side of the transformer the condition is similar to that which happens when the fault occurs on the primary side. The obtained results related to the implementation of the detection method in the cases of inter-turn faults on various phases of the secondary winding have been shown in Fig.6, to Fig.8,. More obtained results related to the turn-to-turn faults have been presented in TABLE II, and Fig.9,, to demonstrate good performance of the protection scheme for detection and localization of the inter-turn short circuits. As can be seen from TABLE II, in the entire of internal fault cases, the amplitude of negative sequence currents are greater than 0.01 pu and also the fault detector is equal to %, so the proposed technique has performed the fault detection process, successfully. According to Fig.9, a pattern similar to Fig.4, can be drawn for these fault cases and determine the faulty phase. Therefore, the protection method has a good performance in the fault localization as well as fault detection. 4

5 Fig. 5. Performance of proposed algorithm for a turn-to-turn fault involving of turns on the phase C of the primary winding: negative sequence Fig. 7. Performance of proposed algorithm for a turn-to-turn fault involving of turns on the phase B of the secondary winding: negative sequence Fig. 6. Performance of proposed algorithm for a turn-to-turn fault involving of turns on the phase A of the secondary winding: negative sequence Fig. 8. Performance of proposed algorithm for a turn-to-turn fault involving of turns on the phase C of the secondary winding: negative sequence 5

6 R TABLE II. PERFORMANCE OF THE PROPOSED SCHEME UNDER VARIOUS INTERNAL FAULT CONDITIONS Fault detector (%) Amplitude of II ss nn (pu) Amplitude of II pp nn (pu) Transformer condition Phase A Phase B Phase C Phase A Phase B Phase C Turn-to-turn fault in primary winding Turn-to-turn fault in secondary winding transformer outside the transformer protection zone and performance of the algorithm in these cases has been shown in Fig. 10, and Fig. 11,. Fig. 10. Performance of proposed algorithm for a phase to phase external fault: negative sequence Fig. 9. Negative sequence current phasor related to the primary side of the transformer in the case of turn-to-turn faults in: primary winding; secondary winding. B. External faults The presented results in this section are used to demonstrate stability of the protection scheme under external faults. The phase to phase (A-C) and phase to ground (A-G) faults have been simulated on the secondary side of the Fig. 11. Performance of proposed algorithm for a phase to ground external fault: negative sequence 6

7 It is observed that the negative sequence currents from both sides of the transformer are nearly equal in magnitude and phase after the fault occurrence, so that their waveforms overlap thoroughly. The external faults lead to severe asymmetry in the system due to flow the great fault currents that cause significant increment in the negative sequence currents. However, the fault detector value remains below 1% threshold. This indicates that the fault is outside the transformer protection zone and hence, no trip command will be issued. C. Load imbalance In this section, performance of the proposed technique has been verified in the case of load imbalance. In order to simulate a three phase imbalanced load, in the disturbance moment, the resistive load connected to the phase A and C have been decreased from 1.6 Ω ( rated value) to 1. Ω and 1.4 Ω, respectively. The obtained results for this case have been shown in Fig. 1, that are similar to corresponding results of the external faults. Fig. 1. Performance of proposed algorithm for the case of load imbalance: negative sequence V. CONCLUSION This paper described a new negative sequence current based protection method for detecting and locating of interturn faults in transformer windings that conquers the limitation of the traditional transformer protection schemes in detecting low level inter-turn faults. The evaluation of the scheme had been done for different faults and operating conditions. It is found that the proposed scheme can not only detect and locate minor inter-turn short circuits, but also discriminate, with a high degree of dependability, between an asymmetry due to an inter-turn fault and an imbalance caused by load and external fault conditions. Hence, the method is completely robust in such external disturbance cases. It can be claimed that the proposed algorithm is very comprehensive because of using the principle of ampere-turns balance in the negative sequence network of the transformer and it is possible to define the algorithm for any type of transformer connections. Since the proposed method only needs the terminal current data, it is non-invasive method and no additional measurements are required to implement it. Also, no information concerning the transformer and power system parameters is needed for the application of the technique. REFERENCES [1] K. L. Butler-Purry and M. Bagriyanik, Characterization of transients in transformers using Discrete Wavelet Transforms, IEEE Transaction on Power Systems, vol. 18,pp , 003. [] K. L. Butler-Perry and M. Bagriyanik, Identifying transformer incipient events for maintaining distribution system reliability, in Proceeding the 36th Annual International Conference on System Sciences, Hawaii, 003. [3] A. Ashrafian, M. Rostami, G. B. Gharehpetian, and S. S. Shafiee Bahnamiri, Improving transformer protection by detecting internal incipient faults, International Journal of Computer and Electrical Engineering, vol. 4, pp , 01. [4] Q. Zhang, Sh. Jiao and Sh. Wang, Identification inrush current and internal faults of transformer based on Hyperbolic S-transform, ICIEA, 009. [5] R. S. Bhide, M. S. S. Srinivas, A. Banerjee and R. Somakumar, Analysis of winding inter-turn fault in transformer: a review and transformer models, IEEE ICSET, 010. [6] T. S. Sidhu, H. S. Gill and M. S. Sachdev, A transformer protection technique with immunity to CT saturation and ratio-mismatch conditions, IEEE Canadian Conference on Electrical and Computer Engineering, vol. 1, Waterloo, Ontario, Canada, [7] T. S. Sidhu, H. S. Gill and M. S. Sachdev, A numerical technique based on symmetrical components for protecting three-winding transformers, Electric Power System Research 54: 19 8, 000. [8] H. Khorashadi-Zadeh and Z. Li, A sensitive ANN based differential relay for transformer protection with security against CT saturation and tap changer operation, Turk J Elec Engin, vol. 15, 007. [9] M. M. Eissa, E. H. Shehab-Eldin, M. E. Masoud and A. S. Abd-Elatif, Digital technique for power transformer fault detection based on positive sequence admittance approach, IEEE 008. [10] V. Mishra, S. Prakash and A. Singh, Detection of internal faults in transformers by negative sequence current, S-JPSET, vol. 4, pp , [11] Z. Gajic, I. Brncic, B. Hillstrom and I. Ivankovic, Sensitive turn-to-turn fault protection for power transformers, CIGRE, Study Committee B5 Colloquium, Calgary, Canada, 005. [1] Mariya Babiy, turn-to-turn fault detection in transformers using negative sequence currents, A Thesis Submitted to the College of Graduate Studies and Research In Partial Fulfillment of the Requirements For the Degree of Master of Science, Department of Electrical and Computer Engineering University of Saskatchewan, 010. [13] A. Vahedi and V. Behjat, Online monitoring of power transformers for detection of internal winding short circuit faults using negative sequence analysis, European Transactions on Electrical Power 1: ,