online partial discharge monitoring and discharge ......often lead to more profit oriented...

1

Transformer Life Management Conference 2014/ Neuss/ Germany

Online Partial Discharge Monitoring and discharge localization on transformers by means of UHF method

J. Pearson, E. Morales, C. Brown, F. Cook, T. Linn; Qualitrol Company LLC; M. D. Judd, M. Zhu; High Voltage Technologies Research Group, University of Strathclyde, Glasgow, UK

Abstract - The electrical industry has experienced a change in direction to intelligent electrical energy network technologies and new requirements for stability and reliability of key network components, over the last 10 to 15 years. This development will continue in the future. Furthermore the developed markets like Europe, US and Japan are facing an aging fleet of key network components, which should be economical operated, without breakdowns till end of their operational life. Additional to that, refurbishment programs based on results of diagnostic and monitoring are being developed to overhaul aged parts and to extend the life of the key components. In addition to that, the deregulation of the electrical energy market forced a higher attention to the cost factors.

Transformers are one of the key components in the network. As such, transformer monitoring parameters like Partial Discharge (PD) are gaining more importance. Conventional PD monitoring is being used, mainly through bushing taps, but suffers from determining the origin of a detected discharge. Newer approaches like UHF PD monitoring are becoming state of the art.

The UHF signal, transmitted by a fast discharge impulse will be picked up by a UHF antenna (sensor – disc like plate, mounted to a transformer). UHF sensors for transformers are intrusive and can be integrated, installed best during engineering and transformer manufacturing. The advantages of the UHF method are the noise immunity and high sensitivity. Furthermore by installing several sensors (4 to 6) a localization of the PD source is possible. Even thought, this unconventional PD measurement method cannot be calibrated, localization and sensitivity verification would allow the user to estimate a magnitude range of a PD source in pC.

I. INTRODUCTION The face of electrical energy production, transmission and distribution has changed significantly over the last few decades. The deregulation of the energy market, along with the privatization of the before public owned utilities, has often lead to more profit oriented enterprises. Furthermore power generation, transmission and distribution were separated in different divisions and privatized separately. Under pressure to increase profits and efficiency, it was popular to outsource maintenance and other technical services. The new created generation, transmission and distribution companies carefully selected there investments in new equipment or in the renewal of equipment. Investments are sometimes limited to the replacement of out of date or failed assets. Even failed equipment has not been replaced in some cases, solely due to pure financial

reasons. The distribution and transmission companies were especially affected by increasing prices from the major power generation utilities and much lower prices in the retail markets, for example in 2000 and 2001 in California (1). The result was the weakening of the electrical network, in a number of cases.

Today after overcoming these teething troubles by introducing additional measures (US in 2002 and in the EU in 2007), private investment in the energy sector has led to the development of efficient equipment and efficient methods for operating the assets, assessing the condition of major network components to maintain the ability to deliver electrical energy and to use the equipment till its real end of live. This is today an important driver for the rapid development of innovative condition monitoring.

Additionally, the old economies are faced with aging key network components. Continuous monitoring solutions were not used extensively in the past, due to n-1 availability of the main equipment. Furthermore, monitoring solutions were less reliable or not available 10/ 20 years ago. Nevertheless, the condition assessment in the utilities of the old economies was carried out under comprehensive periodic measurements and maintenance programs. Nowadays these utilities are seeking more and more solutions to operate their costly components till the real end of life and condition monitoring and condition based maintenance are seen as an important factor to achieve this. However, for almost all of the aged assets no sensors for monitoring had been integrated during its original installation (e.g. UHF sensors for GIS and Transformer, fiber optic temperature sensors etc.). Nevertheless in lots of cases sensors can be retrofitted today.

The change in global energy politics has driven the electrical power industry not only for more efficient solutions, but also to use renewable energy resources, like wind power, geothermic power and solar power. The energy production will become more and more decentralized. Sometimes the energy will be generated far away from the consumption centers, which is the case in terms of offshore wind farms (e.g. in the North Sea). The decentralized power generation leads into a reconfiguring especially of the transmission network. The electrical energy now needs to be transmitted from the regions, where it is generated to the load centers. The control of the decentralized network will be taken over by Smart Grid technologies. Due to the permanently changing load flows, the impact of failing major equipment can only be analyzed by complex simulation. Even the importance of the key components can change with the change of direction of the load flow, which is difficult to integrate into automated reliability and profitability

2


calculations/ simulations. In this respect the collection of online condition data is essential.

Often in terms of offshore installations, the equipment is difficult to access and the space is limited. The installation of redundant equipment is not possible. Furthermore the offshore equipment is exposed to a harsh environment. In these cases the equipment can only be observed remotely.

The new economies, especially the BRIC countries (Brazil, Russia, India and China) are today trying to cover their rapidly increasing energy demand. Therefore the investments in the power industry are immense, which is giving these countries the opportunity to apply the newest and most efficient technologies. In many cases online monitoring and modern condition based maintenance principles are applied or going to be applied in the future. For new assets, it is an advantage to have sensors installed during manufacture.

Due to the above described changes in the technical and political environment, condition monitoring of key assets is gaining more and more importance.

Transformers are one of the key assets of electrical networks and enjoying based on the above described circumstances a high attention in regards to condition assessment. Offline condition assessment methods are established and used since decades with success. Nevertheless, they are giving a screenshot of the transformer in that moment, when the measurements were taken and tests were done. Despite this, the development of the health condition can only be estimated. The development of incipient faults can be missed.

Nowadays besides offline methods, more online monitoring approaches for the transformer fleets are also used in order to capture changing conditions in real time. Among others, partial discharge activity is one of the earliest warning parameter for incipient faults.

Although assessment parameters are proposed in (2), the conventional method under online conditions (PD monitoring on bushing tap/ using bushing as a coupling capacitor according to IEC 60270) solely suffers from the limited ability to distinguish between external discharges (external corona etc.) and internal discharges. The robustness against external noise is also limited. Furthermore a localizing of the discharge inside of the transformer is not possible.

In the late eighties of the last century, UHF PD monitoring for Gas Insulated Switchgear (GIS) was developed and used successfully for more than twenty years. As transformers per design are fully metal encapsulated (as GIS is too), the first efforts to use this method for transformers as well were started in the beginning of this century. The first real installation was done in 2006 for a phase shift transformer in Far East, at that time integrated into the monitoring system of a GIS.

Nowadays UHF PD monitoring for transformers is known and more and more accepted. Nevertheless, there are no general sensitivity or amplitude verification methods throughout the industry agreed and commonly accepted, although there are similar methods as for GIS installations applied today already. One major concern is the fact, that the UHF method cannot be calibrated in pC, especially for a complex structure like a transformer.

The following chapters will explain the basic principle of the UHF method and will present the status of the developments and researches to overcome the above described drawbacks.

II. UHF PD DETECTION PRINCIPLES IN TRANSFORMERS

Electrical impulses caused by partial discharges, have an order of less than 100ps rise time and those produce RF signals extending to several gigahertz. The wave propagation, the original PD pulse shape and the output signal at the UHF sensor is shown Figure 1. The partial discharge source is acting like an antenna and emitting these high frequency signals into the environment. The high frequency waves are traveling and experiencing attenuation and reflections along its way. An antenna inside of the electromagnetic wave field will receive this electromagnetic waves and will provide a signal output for further signal processing.

As UHF sensors, flat metallic plates with a certain shape and a certain diameter can be used. Preferable these sensors are being installed during manufacturing (special flanges are needed to be added to the transformer tank wall) or being installed as retrofit (drain valve or hatch cover sensors) – see Figure 2.

Figure 1: UHF wave generation by PD impulse

3


Figure 3: Examples for Sensor positions on transformer tank

Figure 4: Sensor retrofit example and installation procedure

Figure 3 is showing a sensor placement on a new transformer and Figure 4 and example for a sensor retrofit, using the hatch covers. In Figure 4 the installation procedure is roughly described.

Considering the UHF PD measurement, the closer the PD source is to the sensor, the bigger the output signal is at the sensor, which is different to the signal behavior of the conventional PD measurement according to IEC 60270. This is the result of the attenuation along the propagation way. Also the captured signal shape is not identical with the original impulse, which is due to the multiple reflections and overlapping of the waves inside of the transformer tank according to high frequency signal behavior.

Assuming a certain discharge position and a certain sensor position the following two statements are always true:

• A higher PD intensity is creating a higher UHF output signal at the sensor.

• The travel time from the PD source to the sensor is always the same.

An additional factor to assess the condition of a transformer is the type of defect. According to (2) the differentiation is between the following defects:

• Sharp point causing void. • Sharp point embedded in insulator. • Punctured paper increasing stress. • Loose paper causing void. • Winding movement. • Conductor to conductor discharge. • Surface discharge. • Void adjacent to insulator. • Void on conductor surface. • Corona gas discharge

The differentiation is based on the partial discharge patterns, witch according to the experience are identical for the conventional IEC 60270 method and the UHF method.

III. UHF MAGNITUDE TO PC CONVERSION According to (3) the defect classification in terms of its magnitude is described as following:

Defect free 10-50 pC Normal deterioration <500 pC Questionable 500 -1000 pC Defective condition 1 000-2500 pC Faulty (Irreversible) >2500 pC Critical >100,000-1,000,000 pC The most asked question is how to access the values by the use of the UHF method?

In order to find an answer to that question, several measurements with different transformer models were carried out. All of the measurements are done at

Figure 2: Different types of UHF transformer sensors

4


Strathclyde University by the team of Prof. Martin Judd as a joined project between Strathclyde University and Qualitrol DMS. All of the results are summarized in (4).

In order to create real PD impulses, a test cell with an adjustable gap was used (see Figure 5). By changing the gap, the discharge magnitude and the inception voltage could be changed. The generated discharge is independent from transformer/ test chamber/ transformer model. The voltage needed to create PD depends on the gap distance and is relatively low and can be supplied separately, without energizing the whole transformer/ transformer model (see Table 1).

Figure 5: PD test cell

Gap Size / mm PD Inception Voltage / kV 0.05 4.8 0.10 7.3 0.15 8.8 0.20 10.9

Table 1: PD inception voltage

To prove the PD pulse repeatability the test cell was placed in an empty quadratic test chamber with an edge length of 1m and a UHF sensor installed in the chamber wall.

Figure 6: PD pulse distribution (conventional and UHF method)

As shown in Figure 6 (even the left histogram is more stretched) both histograms are looking quite similar. Additionally it can be seen, that the discharge magnitude is fluctuating between 100 and 600 pC for the same discharge source. The calculated mean value is 417pC, as the mean value for the UHF measurement is 50.5% scale. Just in an empty test chamber, the PD pulses measured by the conventional method and by the UHF method are correlating quite well. It can be concluded, that also in a real transformer tank, PD impulses from the same source, detected at the same UHF sensor are correlating too.

Figure 7: PD pulse distribution of all four installed sensors

Figure 7 is showing the pulse distribution for four different UHF sensors installed in a more realistic transformer model (Figure 8). Also these results are showing a similar distribution of the PD impulses as recorded for the empty test chamber.

Figure 8: Transformer model with winding and core model

The next question, which needs to be answered, is: How to convert the measured mV/ dBm/ Scale percentage to a pC value?

5


Figure 9 is showing different partial discharge magnitudes compared with the UHF signal output. A nonlinear relationship can be discovered, but corresponding ranges between UHF output and apparent charge for that certain configuration can be assumed. Taking that into consideration, the following statement can be made:

If the location of the source is known and sensitivity/ amplitude verification was done, the estimation of the magnitude of the apparent charge based on the range of the UHF output (mV/ dBm/ % of scale) can be done.

Figure 9: IEC apparent charge vs. UHF output

Considering the statement above, it can be concluded the following: To determine the range of IEC 60270 discharge magnitude and to assess the condition of the transformer regarding PD by the reading of the UHF PD magnitude, that:

- The location of the PD source needs to be known - The type of defect needs to be known - The sensitivity/ amplitude verification similar to the

proposed method by CIGRE for GIS needs to be applied.

- The propagation way of the UHF waves needs to be known.

The location can be assessed by means of triangulation. With the complex construction of a transformer, it will need more than only 3 sensors, in order to reach a satisfying accuracy. Localization will be discussed more in detail in the next section.

The type of defect can be assessed by using phase resolved patterns, which are similar for the conventional and the UHF method.

The application of the sensitivity/ amplitude verification today is not common practice and not commonly defined. Therefore further development and discussions in that area are needed.

Out of the design information/ drawings, the propagation way can be assessed.

IV. PD SOURCE LOCALIZATION BY MEANS OF THE UHF METHOD

The localization of a PD source by means of the UHF method is based on the different travel times of the UHF signal from the source to the sensors. In terms of a complex transformer design, reflections and travel ways along the surface of conducting material has to be considered. Therefore more than only 3 sensors are necessary for an accurate localization. Bigger transformer units would require 6 UHF PD sensors and smaller units 4 UHF PD sensors. The placement of the sensors should be in a way that the sensors are having the maximum possible distance to each other. An example is shown in Figure 10.

Figure 10: Example for sensor placement

The sensors at the back side of the transformer preferable are arranged as an opposite triangle compared to the front side.

With the information of the time difference of arrival of the UHF signal to the different sensor, the estimated localization can be calculated.

6


Figure 11: Example for time difference of arrival

In the below example, the transformer was equipped with four sensors and an impulse was injected at sensor 4. The time differences are shown in Figure 11. The time difference calculation has to consider the design of the transformer (basically to determine between metallic and nonmetallic parts – see Figure 12).

Figure 12: PD location calculation software

In order to measure the time difference of the arrival of the UHF signal at the different sensors a common oscilloscope with 4 channels with a sampling rate of 5Gs/s or higher can be used.

Following an example will be described, were the localization of a PD source was applied. A 132kV/ 25kV single phase industrial transformer was showing high DGA reading. Three UHF sensors were retrofitted into three hatch covers at the top cover of the transformer (Figure 13).

Figure 13: Single phase transformer with 3 hatch cover sensors

The PD was confirmed by detecting by means of the UHF method. As the next step, the time difference of arrival was measured (Figure 14).

Figure 14: Time difference of signal arrival

The transformer model was setup in the localization software (light blue = nonmetallic/ light red metallic parts) and the location was calculated (Figure 15 and Figure 16).

7


Figure 15: Transformer model

Figure 16: PD location display

After core disassembly the position of the discharge could be confirmed (see Figure 17).

Figure 17: Confirmation of discharge location

V. CONCLUSION The UHF technique is effective for detection and localization of partial discharges inside of transformers. The installation of UHF sensors is simple for new transformers. To retrofit UHF sensors to existing transformers, hatch covers can be used, but requires an outage. Drain valve sensors can be installed without outages, but in most of the cases only one drain valve sensor can be installed.

An estimation of the IEC PD magnitude range is possible, but requires localization of the PD source and a sensitivity/ amplitude verification preferable during manufacturing. No common agreed method is available today. Further investigations and discussions are required in that field. In regards to that, the new CIGRE JWG A2/D1.51 was established in August 2014, which is dealing with the non-conventional PD measurement techniques for transformers.

UHF PD monitoring on transformers offers new possibilities to assess continuously the condition of a transformer and support the approach to detect changes in its behavior. Nevertheless further research and development needs to be done to assess the health condition of transformers adequate to the given guideline as per (3).

VI. BIBLIOGRAPHY 1. Beneyto, Sylvia and Vantyghem, Julien. discussion-on-the-deregulation-of-electricity-markets (from Mines Paristech). http://www.energypolicyblog.com. [Online] 06 06, 2010. [Cited: 01 20, 2014.] http://www.energypolicyblog.com/2010/06/06/discussion-on-the-deregulation-of-electricity-markets/.

2. CIGRE TF 15.01.04. Partial Discharges in Transformer Insulation. s.l. : CIGRE Paper, 2000.

3. CIGRE WG A2.18. LIFE MANAGEMENT TECHNIQUES FOR POWER TRANSFORMERS. 2003.

4. Zhu, Minan. Calibrating UHF PD Signals from Power Transformers for Equivalent IEC-60270 Apparent Charge. s.l. : Strathclyde University, 2013.

8


Abstrakt (D): Es ist zu beobachten, dass der Energiesektor weltweit über die letzten 10 – 15 Jahre immer mehr durch die Entwicklung hin zu intelligenten Netzwerktechniken und steigende Ansprüche an die Zuverlässigkeit und Stabilität der Kernkomponenten der elektrischen Netze beeinflusst und getrieben wird. Diese Entwicklung wird auch in den nächsten Jahren anhalten. Zusätzlich sehen sich die hoch industrialisierten Länder, zum Beispiel in Europa, den USA und Japan konfrontiert mit dem möglichst störungsfreien Betrieb einer alternden Flotte der Key- Netzwerk- Komponenten, welche effizient und bis zu ihrem eigentlichen Lebensende betrieben werden sollen. Sanierungs- und Modernisierungsprogramme, basierend auf den Ergebnissen periodischer und kontinuierlicher Messungen, wurden entwickelt um gealterte Teilkomponenten zu erkennen und zu überholen und damit die Lebensdauer der Key- Komponenten zu erhöhen. Weiterhin führte die Deregulierung des Energiemarktes dazu, dass Kostenfaktoren eine immer größere Rolle spielen.

Transformatoren sind eine der Key- Komponente der elektrischen Energieverteilung. Als solche gewinnt die permanente Zustandsüberwachung in Form des Online Monitorings von Parametern wie zum Beispiel Teilentladungen, eine steigende Bedeutung. Das konventionelle Teilentladungs- Monitoring, hauptsächlich unter Nutzung des Messanschlusses der Durchführungen, ist heute allgemein verbreitet. Diese Form des TE Monitorings leidet allerdings an der fehlenden Aussage über die Herkunft der gemessenen Entladungen. Neuere Methoden, wie zum Beispiel die Nutzung der UHF Technologie im TE- Monitoring werden heute auch im Transformatorenbereich zum Standard.

Als Folge von schnellen Entladungsvorgängen werden UHF Signale ausgesendet und können durch eine UHF Antenne (Sensor – Antenne, in einer ähnlichen Form wie eine flache, abgerundete Platte, installiert im Transformator) empfangen werden. UHF Sensoren für Transformatoren ragen in den Transformator hinein und werden am besten schon während des Engineerings und der Produktion des Transformators integriert bzw. installiert. Die Vorteile der UHF Technik sind die hohe Störfestigkeit und die gute Empfindlichkeit. Weiterhin kann beim Einsatz von mehreren Sensoren (4 bis 6) die TE Quelle lokalisiert werden. Auch wenn diese unkonventionelle TE- Messmethode nicht kalibriert werden kann, erlaubt die Lokalisation und eine Art vorgängiger Empfindlichkeitsnachweis die Abschätzung der TE Amplitude in einem Pico Coulomb Bereich.

Thomas Linn Qualitrol Company LLC 1385 Fairport Road, Fairport, NY 14450 USA Tel.: + 41 79 276 69 63 E-Mail: [email protected]

online partial discharge monitoring and discharge ......often lead to more profit oriented...

Documents