DETECTION OF INTERNAL WINDING FAULTS IN POWER TRANSFORMER : MODELING AND ANALYSIS

The power transformer is an important and costly element in electric power systems. The continuity of transformer operation is vital in maintaining stability and in improving the reliability of power systems. Any fault in a power transformer affects the operation of the connected power system. In order to reduce the losses caused by the severe currents and avoid the instability in the power system, the correct and prompt diagnosis of fault in damaged power transformers is of great importance. A new and efficient approach for diagnosing the occurrence of incipient turn to turn short-circuits in the

windings of power transformers is discussed here. The transformer is modeled in steady state and no-load operation. So the phase difference between the input voltage and input current in the faulty primary winding is changed in a manner that makes it a good measure to find the fault.

CHAPTER 1

INTRODUCTION

Electrical power transformer plays a vital role in power transmission. It is an important and expensive element in the electric power system. It is a device that consists of a coil wrapped around an electromagnet that transfers electricity from one circuit to another without changing the frequency of the electrical energy. The availability and longevity of power transformers have a major impact on grid reliability and profitability. Any fault in the power transformer affects the operation of the connected power system.

Therefore correct diagnosis of a fault in a damaged power transformer is essential in order to reduce the losses caused by severe currents. 70 %-80 % of the number of transformer failures lead eventually to internal winding faults. Winding faults are a result of the degradation of the transformer winding due to aging, high voltages, etc., which tend to cause a breakdown in the dielectric strength of the insulation. This breakdown either causes adjacent windings to short (turn to turn) or a winding to be shorted to ground. The induced EMF in shorted turns is the source of current in shorted turns. The induced current will flow in a direction so that its magnetic field opposes the main flux in the core. Then the input current is increased to keep the main flux constant. The impedance of shorted turns is very low; therefore, a very high current will flow in shorted turns. in power transformers, for incipient faults or when the number of shorted turns is less than 10 % of total winding turns, there is no significant change in the terminal current to provide protection. Therefore detection and diagnosis of fewer winding inter-turn faults is a difficult task. These faults can lead to a catastrophic failure and

hence cause outages if they are not detected in the early stages. Conventional protection for internal winding faults in power transformers is the percentage differential relay. However, this protection does not completely protect the transformer for fewer numbers of shorted turns. Dissolved Gas Analysis (DGA) is currently used for detection of incipient faults. DGA can be used to verify the effects of faults even in the early stages of fault development. The results obtained from the DGA, which are applicable to an in-service transformer, do not clearly determine the cause and origin of faults. Also, it is not applicable for the dry-type, naturally cooled (NC), and air-forced (AF) cooling based transformers. FRA (Frequency Response Analysis) is also a well-known and popular method in fault diagnosis in power transformers. This method is used for detecting winding deformation and inter-turn short circuit. Conventional FRA has been relying on a graphical analysis for diagnosis of transformers, which requires trained experts to interpret test results. Hence fault detection by this method especially in the first stages is difficult and controversial.

The method discussed here is based on measurement and variations of the phase difference between the no-load input voltages and currents in primary windings of power transformers. The transformer is modeled in a steady state. In modeling under no-load operation, shorted turns in the faulty phase behave similarly to the secondary of an auto-transformer. A custom-built, 3-kVA, 50-Hz, 380/220-V, 3 phase transformer is designed and constructed in our laboratory to perform a set of experiments. The experimental results show that the proposed method is very efficient and encouraging for the detection of internal faults in the windings of distribution and small power transformers even in the early stage.

CHAPTER 2

DETERIORATING INSULATION MODEL

Electrical properties of dielectric material in power transformers may be altered significantly due to operating conditions and diverse factors,. Deterioration and aging of the insulations in a transformer are usually due to the strong electric field that the transformer is encountered with. Incipient internal winding fault in transformers is initiated by deteriorating the insulation between the turns in the windings. Thermal, electrical and mechanical stresses, moisture, and so on are the other factors which generally affect the aging and deterioration of insulation. Generally, the electrical behavior of dielectric material is modeled as a parallel equivalent circuit as is shown in Fig 2.1. This model will be used to simulate the incipient internal faults in the transformer winding. The resistance Rp represents the power loss in the dielectric. The capacitance Cp in this model is given by, where C0 is the equivalent capacitance in the vacuum of the insulation and ǫr is the relative permittivity or dielectric constant.

The insulation thickness between two layers or two turns in a winding is very thin. So the capacitance in the equivalent circuit has a very small value (about nF). The capacitive reactance hence will be very large around the power frequency and can be ignored. Therefore, a proper model that simulates the dielectric behavior is an external resistance between the turns. The resistance decreases with more deterioration of the insulation properties. In other words, losses for perfect insulation is almost zero,

but with degradation and aging of insulation, these losses are rising which leads to a reduction in insulation resistance and when the insulation was broken down completely, this resistance will be almost zero.

CHAPTER 3

ANALYSIS AND THEORETICAL BACKGROUND

3.1 EQUIVALENT CIRCUIT OF THE TRANSFORMER IN PRESENCE OF SHORT CIRCUIT

If there is an internal fault in one of the primary winding of a power transformer under no load operation, the faulty winding can be considered as the primary of an autotransformer. Short-circuited turns are considered as the secondary winding and the fault impedance acts as an autotransformer load. Magnetic flux induces an EMF in the faulty turns. As the short circuit occurs across a few turns of the winding, current flows in the faulty turns. This current generates a magnetic flux that opposes to the initial magnetic flux of the core. Now the input current in the faulty winding increases so that it compensates the effect of current in the shorted turns. The transformer in presence of a fault in the primary winding is shown in figure 3.1. In this case, the equivalent circuit of the transformer can be considered as shown in figure 3.2.

3.2 THEORETICAL BACKGROUND

Figure 3.3 shows the equivalent circuit of one phase of the transformer under no-load

the condition where there are no shorted turns in its windings.

the input impedance of a healthy transformer neglecting the primary winding impedance

is given by

At no-load condition, the transformer acts as a single winding with high self-inductance so that for most of distribution and power transformers the no-load power factor averages about 0.15 lagging. This means that the angle of input impedance at no-load condition is around 80 degrees and the value of core resistance (Rc) is approximately 6-7 times the value of magnetizing reactance (Xm) (so Xn = 6-7 times Rn). Therefore, the transformer input current lags approximately the input voltage by an angle of 80 degrees. For some extra high voltage transformers, this is not true because the no-load power factor may be around 0.85. Therefore, the analysis and the proposed method of fault diagnosis in this paper are valid for those transformers with lower values of no-load power factors.

3.2.2 Faulty condition

If there is a complete internal short circuit across a few turns in the primary winding, it is difficult to detect and diagnose the fault. This method is used in such cases. the analysis is valid for the case where the magnetic fluxes of the two healthy phases are not altered due to the fault current in shorted turns. At a high number of shorted turns, the magnetic flux distribution is fundamentally altered in a way that the magnetic fluxes of the other two phases are changed and so their no-load currents. 

According to the equivalent circuit in figure 3.3, the input impedance can be evaluated. The actual value of resistance and leakage inductance of shorted turns are very low especially for a low number of shorted turns i.e. one, two or five turns. However, the ratio of the number of turns in the primary winding to the number of turns that is short-circuited is actually high. Therefore the referred values of these quantities to the primary side are so that their combination with Zn=Rn+jXn determine the major part of the input impedance of the faulty phase winding. On the other hand, the impedance of primary winding (Z1 = R1 + jX1) is negligible to both Zn and Zf = Rf + jXf and their series combination. The input impedance is given by,

It is seen that under the faulty condition, the above angle is dependent on Rf and Xf.

If this dependence obeys a systematic rule, it is possible to find a method to detect the

inter-turn fault in a transformer.

The angle of input impedance varies considerably as the number of shorted turns

varies. Suppose that < Zinf is smaller than < Zinh. To satisfy this condition, the

following inequality has to be checked:

If the following inequality is satisfied, then the angle of input impedance in the faulty condition is decreased compared to the normal condition. Any increase in the number of shorted turns leads to more decrease in the angle of input impedance.

CHAPTER 4

EXPERIMENTAL RESULTS

A three-phase, YY, core type, 3-kVA, the 380-V/220-V transformer was designed and constructed for the experiments. The primary winding has 270 turns and the secondary has 150 turns. Short circuit and open circuit tests are performed to obtain the equivalent circuit parameters of the transformer. The results are shown in Table 4.1.Table 4.1: Calculated Phase Difference Between the Fundamental Components of the input Voltage and the Input Current in R- Phase Under Different Conditions (Y-Y

Connection)

To establish the short circuit between the primary turns, additional wires are soldered to the selected turns, 250th, 255th, 257th and 258th turn in the primary winding. Wires are extracted out to create the short circuit faults in the primary winding. It should be noticed that the resistance of these wires, acts as the fault resistance Rfault.

Angles of input impedance are calculated in the normal and in the faulty conditions, separately using the equivalent circuit obtained for the laboratory transformer. In the experiments, the short circuit is established through the external wires. The resistance of external wires has also been considered in the calculations. Table 4.1 shows the resistance of external wires and the phase difference between the fundamental components of R-phase voltage and R-phase current under normal and faulty conditions.

The results in Table 4.1 show that as the number of shorted turns increases, the angle of input impedance or the phase difference between the fundamental components of phase voltage and phase current in faulty phase decreases considerably.

The experimental investigation was carried out to verify the presented model and theoretical bases. Steady-state input voltages and currents were analyzed. High voltage windings were connected to the 380-V line to line voltage and experiments are performed under no load condition. Waveforms of the input voltage and the input current in Rphase (faulty phase) are shown in Fig. 4.1, 4.2, 4.3, 4.4. Per-unit values of voltages and currents are plotted. It is seen that as the fault occurs and expands in

the primary winding, the phase difference of the input voltage and the input current decreases significantly in the faulty winding. Experimental results show that the phase difference between the input voltage and the input current in the other two healthy windings has no significant variation while the number of shorted turns is low. Table 4.2 gives the phase difference between the fundamental components of Rphase voltage and R-phase current under normal and faulty conditions. These results have a very close agreement with the calculated results shown in Table 4.1.

Small differences between the calculated and experimental results are due to some unbalances in applied line voltages.

Alternatively, instead of the measurement of phase difference between the fundamental components, we can measure the phase difference between the overall input voltage and the overall input current in the faulty phase. Our investigation shows that the error due to this approximation is negligible. However, it is much easier to measure the phase difference using the overall signals and not their fundamentals. Summary of the results using the overall signals is given in Table 4.3. To confirm and to show the effectiveness of the analytical approach and this method for fault detection for various winding connections, the primary side of the laboratory transformer is changed to delta connection. The faulty winding is connected between phase R and phase S of the three-phase power supply. We name the faulty winding as RS winding. Table VIII gives the phase difference between the overall input phase voltage and the overall input phase current in the R and S phases, respectively. It is seen that as the fault occurs and expands in RS winding, the phase difference between the input phase voltage and the input phase current in phase R and phase S decrease significantly. Experimental results show that the phase difference between the input phase voltage and the input phase current in phase T has no significant variation. The faulty winding has no direct connection to Phase T. So the experimental results show that the method is applicable to DY connection too.

CHAPTER 5

CONCLUSION

Under steady state, no-load operation power factor of a power transformer is approximately very low. Hence its behavior is similar to a winding with high inductance. If there is any turn to turn short circuit in one of the primary windings, this winding and short-circuited turns can be considered as the primary and secondary windings of an autotransformer, respectively. Magnetic flux induces an EMF in the faulty turns. As the short circuit occurs across these turns, current flows in the faulty turns. This current generates a magnetic flux that opposes to the initial magnetic flux of the core. Now the input current to the faulty winding increases to compensate for the effect of current in the short-circuited turns. For internal faults with a low number of turns, the equivalent impedance of faulty turns is mostly resistive. So the current through the faulty winding is approximately in phase with its applied voltage. This method is applicable to various winding connections Y or D. But it is suitable only for the no-load operation of transformers.

REFERENCES

[1] Nahid Asadi, Homayoun Meshgin Kelk, “Modeling, Analysis, and Detection of Internal Winding Faults in Power Transformers ”, IEEE Transactions on PowerDelivery, September 2015.

[2] Karen L. Butler and Adedayo Kuforiji, and D. P. Kothari, “Experimental results from short-circuiting faults on a distribution transformer ”, 2008 IEEE Transmission and Distribution Conf., vol.1, Oct. 2001, pp. 299-306

[3] J.C. Meza and A. H. Samra, “Zero-sequence harmonics current minimization using zero-blocking reactor and zig-zag transformer”, IEEE DRPT, pp. 1758-1764,

2008.

[4] C. E. Lin, J. M. Ling, and C. L. Huang, “An expert system for transformer fault diagnosis using dissolved gas analysis”, IEEE Trans. Power Deliveryvol. 8, no. 1, pp. 231-238, January 1993.

[5] E.P. Dick and C.C. Erven “Transformer diagnostic testing by frequency response analysis”, IEEE Trans. on Power Apparatus and Systems, vol. PAS-97,

no. 6, pp. 2144-2153, Nov. 1978.

[6] Jong-Wook Kim, Byung-Koo Park, Seung Cheol Jeong “Fault diagnosis of a power

transformer using an improved frequency-response analysis”, IEEE Trans.

Power Delivery ., vol. 20, no. 1, pp. 169 178, January 2005