Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright °

ISBN 978-0-620-44584-9

EVALUATION OF ACOUSTIC EMISSION SIGNATURES OF ON-LOAD TAP-CHANGERS M.B. Trindade1* and H.J.A. Martins1 Electric Power Research Center, CEPEL, Av. Horácio de Macedo 254, Cidade Universitária, Ilha do Fundão, Rio de Janeiro, Brazil *Email: [email protected] 1

Abstract: In this work, are presented results of the development of a methodology that, using the acoustic emission technique, aims to verify the conditions of operation and diagnosis of on-load tap-changers. Acoustic signatures obtained from on-site tests, in different models from four manufacturers of tap-changers, are analyzed. The developed method showed to be effective to the desired purpose, with capacity to enable the acquisition of acoustic signatures, performing as an appropriate tool for the diagnosis of on load tap-changers, allowing the discrimination between states of normal operation and with the presence of defects. 1.

equipment. In a normal sequence operation briefly occur the following events:

INTRODUCTION

On-load tap-changers (OLTC) are machines with mechanical behavior following well-defined standards. This characteristic is resulted from the need that the events involved on tap-changer operations must be processed with high precision and synchronism in order to avoid risk to the integrity of power transformers. Since these are mechanical switching devices, on-load tap-changers are expensive and vulnerable devices of power transformers. The largest part of the defects is mainly related to mechanical components and presents higher occurrence ratio for contacts mechanism [1-2].

• • • • • •

Each of these events is processed by mechanical devices, which when operating act as acoustic emission sources, generating signals that can be detected by the measurement system.

OLTC can be designed as a single unit for single and three-phase applications. Depending on the three-phase rating, it might require three separate units, each having its own insulated phases. Nowadays, there are different OLTC models in operation, which vary according to the manufacturer and application. In general, these models can be fit in two basic types [3]: •

•

On load tap changer, Figure 1 – generally, it consists of a diverter switch with a transition impedance and a tap selector which can be with or without a change-over selector. Selector switch – a device combining the duties of a tap selector and a diverter switch. The tap selection and the tap changing are processed simultaneously with the use of the same set of movable contacts.

Besides the type, other characteristics differentiate the OLTC models: • • •

Driving mechanism operation (motor, transmission shaft, gears); Set in motion the change-over selector and change the position of its contacts; Spring energy accumulator loading and releasing; Set in motion the tap selector shaft; Tap selector contacts transition between taps; Diverter switch operation.

Figure 1: On load tap-changer

Insulation: oil (conventional technology) or vacuum (vacuum switching technology); Assembly in the transformer: it can be located either inside the transformer tank (internal) or outside in its own compartment (external); Transition impedance: reactor or resistor.

Acoustic emission is a non-destructive and noninvasive technique useful in detecting internal active defects in materials and equipment [4-5]. Due to their convenience, acoustic emission tests are an important tool of preventive inspection, and are an alternative to other more expensive non-destructive test methods. Previous researches have already indicated it as adequate to support diagnosis of OLTC [6].

During the short period of a tap-changer switching, automatically occur events whose precision and synchronism depend on the right working of the Pg. 1

Paper D-31

Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright °

ISBN 978-0-620-44584-9

the difference in the signatures provoked by the change in sensor position.

Acoustic signals, generated inside the equipment in operation or when subjected to programmed strengths, carry with them important information’s related to their integrity. These signals, generally associated to presence of defects or to typical performing characteristics of the equipment and of its components, can be picked up externally, using acoustic sensors, and analyzed considering the variation in parameters such as: amplitude, energy, counts, duration and average frequency. Studies have revealed that, among several relations between these parameters, the curves of accumulated energy (AcE), during the tap change time, are the acoustic signatures that best characterize the OLTC behavior [6].

The environmental conditions during the tests were considered satisfactory. However, in the case of tapchanger D, the presence of noise, originated inside the power transformer, has interfered in the results. In none of the results, the oil filter was activated. The quantity of switchings, for each equipment, was 24, corresponding to a rising up to 3 positions above the normal position operation, one fall up to 3 positions down and other rising, returning to initial position. Following, the cycle was repeated one more time. Additionally, it were included results obtained from a three-phase tap-changer, type selector switch, which was subjected to monitoring before and after maintenance, in the condition – de-energized, and in which it was identified the presence of defects.

The energy of acoustic signals, informed by the measuring system during the tests, refers to the measured area under the rectified signal envelope (MARSE), in amplitude versus time coordinates [5]. This parameter, for the form as it is obtained consists in a dimensionless measure, being only indicated as units of energy.

3.

RESULTS

3.1. 2.

PROCEDURE

Figure 2 shows a picture framed with typical signatures obtained from switchings performed in different groups of tested OLTC. In this figure, are showed two curves for each switching obtained from two cycles of switchings performed in sequence.

The results showed in this report were obtained in tests performed in 14 one-phase OLTC, from different manufacturers and models which characteristics are showed in Table 1. The OLTC were coded according to the manufacturer (A, B, C and D), the model (1 and 2) and the connected sensor (S1, S2 and S3). Each group of OLTC, of the same model and manufacturer was installed in the same power transformer bank.

In these graphs, the marked dots correspond to the events involved in the switchings. The interconnection of these dots, utilized during curves construction, only attends for better visualization of the accumulated energy values evolution, since the events are discrete. The adopted origin for the time measurements was the instant of the occurrence of the event of major energy. For this reason, some period of time appear indicated with negative values.

Table 1: OLTC characteristics

Manufacturer / model - sensor A1-S1, S2, S3

Diverter switch C

Insulation Assembling Oil

A2-S1, S2, S3

VI

B-S1, S2, S3

VI

C-S1, S2, S3

C

Oil / Vacuum Oil / Vacuum Oil

D-S1, S2

C

Oil

Internal

Transition impedance Resistor

External

Reactor

External

Reactor

Internal

Resistor

Internal

Resistor

Acoustic signatures

The main stages of the switching process can be identified in curves of Figure 2: starting of operation of driving mechanism, tap selection and operation of the diverter switch. The absence of signals between -2000 and 0 ms, observed in Figure 2(b), is explained by the fact that switchings present in this figure are related only to inversion in the direction of the switching, without moving the contacts of the tap selector. This ascertaining comes to confirm the indication of this interval as that in which the tap selection is processed.

NOTE: Conventional diverter switch (C) and Vacuum interrupter (VI)

The tests consisted in monitoring acoustic signals emitted during switchings performed on-site, with the OLTC in normal operation. The monitoring were simultaneously made in each group of transformers. In general, for each tap-changer, it was used only one fixed sensor outside the tank wall. The sensor positions varied according the constructive characteristics of the set tap-changer / transformer, in that way both selection signals and tap change signals were detected with suitable levels of intensity.

The main observations about the curves showed in Figure 2 are: •

• •

In one of the tests, for OLTC B-S3, it was used one more sensor, S4, positioned at the same height of the sensor S3, but in an adjacent side, in the OLTC compartment. This assembly had as objective to verify Pg. 2

The repeatability of the tests verified by the overlapping of the curves, by the repeating of the dots and by the occurrence of time coincidence; The difference between a full switching, Figure 2(a), and a switching without tap selection, Figure 2(b); The difference between the curves generated by the signals detected by sensors in different positions, Figure 2(d). Despite the gap between the energy values of the signals, the shape of Paper D-31

ISBN 978-0-620-44584-9

•

Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright °

•

both curves and the repetition of the dots at the same period of detection are maintained; The difference between a simple switching Figure 2(e), and a switching with intermediate positions, Figure 2(f);

The effects of noises inside the transformer, during the monitoring, Figure 2(g). The interference is more evident when the curves are compared with the curves in Figure 2(h), generated by a similar OLTC, without noises.

(a) Tap-changer A1

(b) Tap-changer A1

(c) Tap-changer A2

(d) Tap-changer B

(e) Tap-changer C

(f) Tap-changer C

(g) Tap-changer D

(h) Tap-changer D

Figure 2: AcE (MARSE) versus time (ms) Typical signatures of tested OLTC In Figure 3, it is possible to notice similar behaviors in the curves of average of the total accumulated energy for switchings in the same group of OLTC. The ripples in some of those curves are generated according to the proximity of the sensor to the taps that are being switched, due to the circular disposition of the fixed contacts of the tap selector. In the specific case of the OLTC C-S1, C-S2 and C-S3, the peaks of total accumulated energy correspond to the switchings that involve intermediate positions and change-over operation.

Values of the average of the total accumulated energy (AcE) for each OLTC in each switching, obtained through the average of the sums of the signals energy in the repetitions, are presented in Figure 3, together with the respective variation coefficient, VC%. In this work, the VC%, calculated by the following formula, was used as a parameter to evaluate the repeatability of results:

VC % = ( s / X ) * 100 where s is the standard deviation and X is the average of values, for each group of results obtained under the same conditions.

In general, the VC% calculated were lower than 5%. The main exceptions were the OLTC D-S1 and D-S2, Pg. 3

Paper D-31

ISBN 978-0-620-44584-9

Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright °

whose VC% reached up to 20%. The first case happened due to the noises generated inside the transformer and that varied during the repetitions, affecting the results of each switching. For the OLTC D-S2, the values suggest a more meticulous investigation.

In Figure 4, the values of the variation coefficients are presented. They were calculated for every switching and every set of OLTC in a same group.

(a) Tap-changer A1

(b) Tap-changer A2

(c) Tap-changer B

Figure 4: VC%, for each switching and group of OLTC The VC% values varied a lot, reaching up to 15%, which may be considered high. This behavior indicates that the quantitative results generated by an OLTC will not necessarily be repeated in another OLTC of the same model and manufacturer.

(d) Tap-changer C

3.2.

Identifying defects

During the maintenance of a three-phase, selector switch type on-load tap changer, one of the fixation screws, placed at upper coupling gear of the selector switch shaft, has been found broken and the others loose, affecting the contact movement. The accumulated energy curves versus time, acquired after and before the repair of this tap-changer, can be observed in Figure 5.

(e) Tap-changer D

The difference of the curves behavior on the two conditions is very obvious and was verified in all tapchanger operations performed. It is possible to observe a significant variation of the accumulated energy values from the moment of the releasing of the spring energy accumulator until the end of the movable arcing contacts transition. These results indicate the presence of a defect that appears at the same form in all tapchanger operations, compatible with the defect identified during a later inspection. The variation of accumulated energy was of 33% at the end of the switching process.

Figure 3: Average of total AcE and VC%, for each switching and each OLTC

Pg. 4

Paper D-31

ISBN 978-0-620-44584-9

Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright °

the results and should be taken under consideration during monitoring.

(a) Tap change signatures

Quantitative comparisons between acoustic signatures should always involve the same equipment, the same switching, the same switching direction and the sensor should be in the same position. The signatures of two tap-changers of the same model can have the same behavior and present the same switching events and periods, but distinct intensities. For the same switchings, the same OLTC and an environment free of noise, the variation coefficients presented in great part values lower than 5%, which indicates good repeatability of the results when compared to the 33% variation in the total accumulated energy, caused by the presence of defects in the selector switching type OLTC.

(b) After spring releasing

With the presented results, is possible to conclude that the methodology under development, using the acoustic emission technique, is a proper tool to assist in the evaluation and diagnosis of on-load tap-changers. 5.

Figure 5: AcE (MARSE) versus time (ms), before and after maintenance 4.

ACKNOWLEDGMENTS

The authors thank Luiz E. D. Santos and Roberto C. de Menezes for the contribution during the execution of this work.

REMARKS AND CONCLUSIONS

6.

The test results show that acoustic emission technique is able to identify the many events involved in the switching process, enabling to evaluate their intensity and the periods when they occur.

REFERENCES

[1] P. Kang, D. Birtwhistle, J. Daley, D. McCulloch, Non-invasive On-line Condition Monitoring of On Load Tap Changers, Proceedings of IEEE Power Engineering Society Winter Meeting, Singapore 2000. [2] CIGRE SC 12 WG 12.05. An International Survey on Failures in Large Power Transformers in Service, ELECTRA, nº88, 1983, pp 21–42. [3] International Electrotechnical Commission. Onload tap-changers, International Standard – IEC 214. [4] American Society for Nondestructive Testing. Acoustic Emission Testing, Nondestructive Testing Handbook, Vol. 5, 2ª Ed. [5] A. A. Pollock, Acoustic Emission Inspection, Metals Handbook, 9ª Ed., Vol. 17, American Society for Metals. [6] M. B. Trindade, H. J. A. Martins, A. F. Cadilhe, J. A. C. Moreira, On-Load Tap-Changer Diagnosis Based on Acoustic Emission Technique, XIVth International Symposium on High Voltage Engineering – ISH/2005, Tsinghua University, Beijing, China, August 25-29, 2005.

The repeatability of the results allowed the establishment of acoustic signatures for on-load tap changers of different models and manufacturers. The methodology developed by CEPEL for obtaining acoustic signatures is non-invasive and easy to apply. It can be used even with the equipment energized and operating. The amount of tap-changers that can be simultaneously monitored is only limited by the number of channels available in the measurement system. The test duration is extremely short, being only the time required for the switchings. The comparison between acoustic signatures obtained before and after the maintenance of a selector switch type three-phase OLTC showed that the defects can be clearly identified by changes introduced into the accumulated energy curves. Noises caused by the operation of the transformer or its components, such as oil filters, besides the noises produced by environmental factors can interfere with

Pg. 5

Paper D-31