2011 Electrical Insulation Conference, Annapolis, Maryland, 5 to 8 June 2011

On-Load Tap Changers Dynamic Resistance Measurement

Settings and Interpretation
J.J. Erbrink, J. Aditya, L.A. Chmura, J.J. Smit
Delft University of Technology Delft, the Netherlands

R. Leich
Liandon Alkmaar, the Netherlands

B. Quak
Seitz Instruments AG Niederrohrdorf, Switzerland

AbstractAn on-load tap changer (OLTC) is prone to many different defects and degradation mechanisms due to its continuous operation under load condition. A high load current and high operation temperature accelerate the process of contact wear. Contacts erode due to the fact that under certain conditions the contacts develop a resistive layer due to hot insulation oil. As a major cause of OLTC failure, these defects might take place after some years of operation and their frequency increases with the operation time. Regular maintenance is applied due to degradation of the insulation oil. Dynamic resistance measurement (DRM) can be conducted during this maintenance to measure the resistance of the OLTC when it operates. The measurement results provide important information for assessing the actual condition of the OLTC. This paper aims to analyze the effect of test parameters on DRM results from service-aged OLTCs and describes the implementation of DRM in OLTC condition assessment. Two test parameters were selected for observation: the test current amplitude and the circuit resistance. Their amplitude was varied during laboratory experiments as well as on-site after regular maintenance. To quantify the various measurement results, an analytical tool using 4 performance parameters has been developed. Finally for interpretation, with the help of a three dimensional classification system which is presented in this paper, the actual condition of OLTCs can be identified during maintenance activities. Keywordsaging, contact resistance, contacts, diagnostic expert systems, dynamic response, power transformers, resistance measurement.

and wear of contacts due to switching arcs. These defects are interesting to analyze because they influence the performance of an OLTC. A resistive layer is formed on the OLTC contacts due to a heating by the load current [4-5]. This results in contact deterioration. Contact wear occurs due to erosion of arcing contacts in the arcing switch [6]. DRM is used to detect such degradation. The results of the measurement are expected to be highly dependent on test parameters, i.e. the test current and circuit resistance. The value of the test current and circuit resistance will be varied during measurements to study the influence of these test parameters. II. EXPERIMENTS

Experiments have been conducted on several selector switch type OLTCs, using dynamic resistance measurements. In previous research it was seen that contact resistance has a different amplitude when measured with different test parameters [7]. A test setup has been built to investigate this effect in more detail and to allow varying the level of different test parameters using a single test setup. A. Test Parameters The test current and circuit resistance are varied in order to study their influence to the DRM measurement results. A low test current has been commonly used in dynamic resistance measurements. It is advised in the IEEE C57.12.90 test code that the test current during resistance testing of power transformers shall not exceed 15% of the rated current of the windings [8]. A higher value may cause inaccuracy by heating the winding. In this research paper, the test current amplitude was subdivided into the following ranges: low (0.2ADC to 0.5ADC), medium (1ADC to 1.5ADC) and high level (5ADC and above). The total circuit resistance of the test circuit influences the dynamic resistance. Therefore it has effect on the sensitivity of the measurement and the amplitude of the measured resistance. Measurements with varied circuit resistance were conducted using a fixed test current of 1ADC. The selected values for the added resistance are low (0.2), medium (1) and high (4). The low value of 0.2 is the lowest value can be reached by the variable resistor that is used.

I.

INTRODUCTION

As the only moving part of the transformers and a major cause of a failure [1], the OLTC benefits from diagnostic measurements performed to observe its condition. One of the diagnostic methods is dynamic resistance measurement (DRM) [2], which is studied in this research. Dynamic resistance measurement is used to diagnose the condition of the OLTC contacts. The result of the measurement is a current or resistance graph. Any deviation in the graph reflects resistance on the tap changer contacts. A defective OLTC can lead to a total power transformer failure and can be caused by different kind of defects on different parts of the OLTC. Ref. [3] described these defects. This paper discusses aged contact due to carbon development

978-1-4577-0276-1/11/$26.00 978-1-4577-0279-2/11/$26.00 2011 IEEE

54

B. Measurement Setup A DRM test setup was developed with the additional tools added to it, see Fig. 1.

healthy condition) due to defects. Parameters P1-P4 are discussed below. A. Parameter 1 (P1): Winding Resistance The linear pattern in the DRM graph is formed by resistance of the transformer windings that are selected or deselected by the OLTC (for example, the winding resistance changes 15% when the OLTC is switched through all taps). This gives the measurement results a pyramidal shape. The amplitude of this pyramid is quantified to determine the sensitivity of the test circuit to changes in the resistance, and is expressed as this performance parameter P1. Based on the charging process and the sequence of tap changing, the amplitude of the graph is calculated in different ways. A measurement performed with a stable test current will result in a graph which has equal value of i1 and i2 (as shown in Fig. 2). Then the calculation of P1 will be:
P1 () = i 3 i1 100% i1

(1)

Figure 1. Test setup for dynamic resistance measurement. The scheme of the circuit connection for Y transformer windings is shown in the upper figure and a picture of the test circuit below.

When the measurement is performed under unstable charging voltage, then i1 i2, this is corrected:
i3 ( i 2 i1 ) i1 100% if the value i1<i2 (2) P1 () = i1

A DC power supply is used to reach 20ADC test current. It is connected in series with a variable resistor and the transformer with OLTC. A 0.2 shunt resistor was used to measure the current. This shunt resistor showed to be temperature independent. The transformer diagnostic system (TDS, [9]) is used as voltage recorder and is connected parallel with the shunt resistor. III. ANALYTICAL TOOL

i3 + ( i1 i 2 ) i1 100% if the value i1>i2 (3) P1 () = i1 Besides the matter of charging, there is a possibility that the tap changers have a different sequence of taps if the lowest tap has the lowest impedance. This leads to a reversed form of DRM graph shown in Fig. 2. For the calculation, a similar approach is then used:

A systematic method of analyzing the measurement results is developed in this section. Four performance quantities (P1P4) will be used to quantify each measurement result. These quantities will then be used to determine the condition of an OLTC. The resulting analytical tool is described in Fig. 2. These selected performance parameters can reveal the most common OLTC defects.

P1 () =
P1 () =

i1 i3 ( i 2 i1 ) i1
i1 i 3 + ( i1 i 2 ) i1

100% if the value i1<i2 (4)

100% if the value i1>i2 (5)

Figure 2. Four performance parameters (P1-P4) are applied to DRM graphs.

B. Parameter 2 (P2): Contact Resistance Short Term Effect The principle of performance parameter P2 is to measure the presence of a resistive layer on briefly used tap-selector contact surface, which develops in the beginning phase of contact degradation. The deviation in the DRM graph can occur in the moment the tap selector is moved while the current flows through it. The location where the tap selector contacts have some resistive layer is shown in closed up image in Fig. 2. A line is used, indicating the expected value. Then the difference between two trigger points (i5 and i4, see Fig. 2) is calculated:

These performance parameters indicate how much the current deviates from the reference (which is an OLTC in

P2 (%) =

i 4 i5 100% i4

(6)

55

C. Parameter 3 (P3): Contact Resistance Long Term Effect The third performance parameter determines the difference in resistance between change-over selector contacts which are frequently used. Those change-over selector contacts are prone to advanced aging and a long term effect of carbon development. A typical result is one with a part of the graph shifted into lower current value due to the increase of resistance on the change-over selector contacts. Again, a trend-line is used by extending the linear pattern before and after the change-over selector operation. The difference value between the expected line and the measured line (i6 and i7, see Fig. 2) is calculated:

P3 (%) =

i 6 i7 100% i6

(7)

Figure 3. Relation between the test current and the sensitivity of the measurement

D. Parameter 4 (P4): Timing of Transition Resistor(s) The changes in contact timing, transition time and changes in the value of transition resistors during OLTC switching is measured with performance parameter P4. This parameter is also sensitive to the changes in the circuit time constant ( = L/R). A large time constant may not allow the measurement to read the fast current change during the tap transition.

The difference between the two points i8 and i9 (see Fig. 2) is calculated for the entire cycle of operation. This value should be constant. When there is any wear of contacts, the value will show fluctuations. Finally, the average of those results will be calculated to determine the sensitivity of the test circuit to wear and erosion:

1 n i i P4 (%) = 8 9 100% n i =1 i8
IV. RESULTS

(8)

Figure 4. The influence of test current to the analysis of short term aging effect

Analysis of the performed measurements is done using the analytical tool. The performance parameters P1-P4 will show the influence of test parameters and degraded contacts to the measurement results. First, experiments with the test current amplitude will be discussed.
A. Test Current The sensitivity of measurement (P1) is independent from test current. Good and light aged contacts show a constant sensitivity, see Fig. 3. However, the graph in Fig. 3 indicates that the medium and low test levels (<1.5ADC) have not broken the thin contact degradation and the degradation still shows up in the low current measurement results. A measurement with high test current broke the resistive layer and restored good contact.

Fig. 5 shows that the influence of the test current to measurement results on long term aged OLTCs (represented by P3) is similar to short term aging (P2). The high test current causes a decreasing contact resistance. It can be seen that good OLTCs have the lowest resistance, and it increases along with aging.

The influence of performance parameter P2 to the test current can be seen in Fig. 4. The measurements, performed on four different OLTC aging phases, show that P2 increases as the OLTC ages further. Another effect of test current can be seen: the sensitivity decreases when the test current is increased. Low current measurements show to be very sensitive to light aging (which is characterized by the value of parameter P2).

Figure 5. The influence of the test current to the analysis of long term aging effect

56

Any variation of current does not have much influence to the measurement results of transition resistor timing (P4). The curves shown in Fig. 6 are considered flat.

Figure 8. The influence of circuit resistance to the analysis of short term aging effect

Figure 6. The influence of test current to the reading of OLTC switching time

B. Circuit Resistance The sensitivity of the measurement (P1) depends heavily on the circuit resistance. Fig. 7 shows that the sensitivity decreases when the circuit resistance is raised. A high circuit resistance produces low amplitude response signal.

Similar results are measured on frequently used contacts (P3). The measurements on good and light aged change-over selector contacts have flat curves, as shown in Fig. 9. Again it is seen that the value of P3 increases as the OLTC ages further, which is best measured at low circuit resistance.

Figure 9. The influence of circuit resistance to the analysis of long term aging effect

Figure 7. Relation between the circuit resistance and the sensitivity of the measurement

The influence of circuit resistance on measuring the effect of transition resistors can be seen in the Fig. 10. Performance parameter P4 decreases linearly when the circuit resistance is tuned higher.

Influence of circuit resistance to the measurement results of short term effects (P2) can be seen from Fig. 8. The graph indicates that advance aging has the highest resistance. Performance parameter P2 is low when the contacts are in good condition and vice versa. The high circuit resistance produces a weaker, low amplitude response signal P2. The presence of a very light resistive layer remains undetected when the circuit resistance has reached 2. When it reached high value of 4, all measurements on aged OLTC (light, medium and advanced phase) tend to have a similar value of P2. It indicates that at high circuit resistance is almost impossible to distinguish between the aging phases.
Figure 10. The influence of circuit resistance to the reading of OLTC switching time

57

V.

THE IMPLEMENTATION ON CONDITION ASSESSMENT

The performance parameters discussed earlier can also be utilized for condition assessment. The performance parameters P1-P4 can then be used to make a decision whether the power transformer can be put back into service or shall have a more extensive overhaul.
A. Selection of the Optimal Test Parameter First, an optimal set of test parameters needs to be selected to perform the measurements. It shall have a balance between the sensitivity to the resistance change and the strength of the measurement signal. It has been determined through the experiments that medium level of test current (1ADC up to 1.5ADC) with minimum circuit resistance is the most appropriate level. A lower current level shows to be too sensitive to medium aged OLTCs. B. Concept of a Three Dimensional Classification Systems In order to simplify the condition assessment of OLTCs, a three dimensional classification system is introduced (Fig. 11). Three performance parameters (P2, P3 and P4) which have been used for OLTC defect analysis are used as the axis. Performance parameter P1 is excluded since it is used to determine the sensitivity of the measurement setup to changes in resistance. Each measurement will result in a 3D coordinate.

OLTCs future condition, based on the trend of the results from different measurements. VI. CONCLUSION

This research discussed the influence of test parameters on dynamic resistance measurement (DRM) results from OLTCs. Two test parameters, which are test current and circuit resistance, were varied during experiments. Four performance quantities were introduced to describe the results. The following findings can be defined: Performance parameter P2 and P3 show a reduced sensitivity when the test current is increased. The effect of test current on P4 is less significant than the other performance parameters. The medium test current has a good balance between the sensitivity and the strength of signals due to noise. It has a reliable sensitivity to defects. The effect of circuit resistance shows similar in all performance parameters: all four quantities reduce when the circuit resistance is increased.

Besides quantifying the measurement results, the analytical tool can be utilized for interpretation of DRM results. A 3D classification system has been developed with purpose to interpret the measurement results and to make prediction of OLTCs future condition. It may simplify the condition assessment of OLTCs. REFERENCES
[1] [2] Cigr WG 12.05. An international survey on failures in large power transformers in service ELECTRA No. 88, pp. 21-48, 1983. H. F. A. Verhaart, "A diagnostic to determine the condition of the contacts of the tap changer in a power transformer," in. Proc. CIRED, vol. paper 1.13, 1995. J.J. Erbrink, E. Gulski, P.P. Seitz, R, Leich, Advanced on-site diagnosis of transformer on-load tap changer, IEEE International Symposium on Electrical Insulation, pp. 252-256, June 2008. J. J. Erbrink, E. Gulski, J. J. Smit, P. P. Seitz, and R. Leich, "Experimental model of aging mechanisms of on-load tap changer contacts," International conference on condition monitoring and diagnosis, Beijing, China, pp. 247-250, April 2008. Z. Wang, Y. Liu, and P. J. Griffin, "Artificial Intelligence in OLTC Fault Diagnosis Using Dissolved Gas-In-Oil Information," IEEE Power Engineering Society Summer Meeting, Seattle, USA, vol. 4, pp. 24222427, July 2000. H. U. Schellhase, R. G. Pollock, A. S. Rao, E. C. Korolenko, and B. Ward, "Load Tap Changers: Investigations of Contacts, Contact Wear and Contact Coking," Proceedings of the Forty-Eighth IEEE Holm Conference on Electrical Contacts, pp. 259-272, 2002. J.J. Erbrink, E. Gulski, J.J. Smit, R. Leich, P.P. Seitz, B. Quak, On-load Tap Changer Diagnosis: Interpretation of Dynamic Resistance Deviations, 2010. IEEE C57.12.90 1987, IEEE Standard Test Code for LiquidImmersed Distribution, Power, and Regulating Transformers. J. J. Erbrink, E. Gulski, J. J. Smit, P. P. Seitz, B. Quak, R. Leich, and R. A. Malewski, On-load tap changer diagnosis an off-line method for detecting degradation and defects: part 1, IEEE Electrical Insulation Magazine, Vol. 26, No. 5, p. 48-58, September-October 2010.

[3]

[4]

[5] Figure 11. The concept of three dimensional classification system. This sample of concept is made for 1.5ADC measurement results.

Each block in the 3D graph represents an aging phase or defect. In order to classify the condition blocks, boundaries were determined. These borders are based on many measurement data from different aged OLTC. The classification of OLTCs into these aging phases is done this way [7]. Fig. 11 is valid only for a selected test level (in this example 1.5ADC) and its boundaries shall be determined based on population data from similar type of OLTCs. With this classification system, the percentages value of P2P4 can be interpreted into the actual condition of OLTCs. Another purpose is that the concept can be used to predict the

[6]

[7]

[8] [9]

58