2013 Annual Report Conference on Electrical Insulation and Dielectric Phenomena
Bridging in Contaminated Transformer Oil under AC, DC and DC Biased AC Electric Field Shekhar Mahmud*1, George Chen1, Igor O. Golosnoy1, Gordon Wilson2 and Paul Jarman2 Electronics and Computer Science, University of Southampton, Southampton, SO17 1BJ, UK 2 National Grid, Warwick Technology Park, Warwick, UK *E-mail:
[email protected]
1
also recorded during experiments for DC and AC electric field.
Abstract- Cellulose particle accumulation under AC, DC and DC biased AC electric field in transformer oil has been investigated in this paper. Different levels of particle concentrations tested with spherical electrode system. Optical microscopic images and conduction current of the particle accumulation process have been recorded during the experiments. A complete bridge of cellulose particles only observed under DC and DC biased AC electric field.
The detailed description of sample tank, preparations, experimental setup is given section II, whereas the results are presented in section III and summary is given in section IV. II. EXPERIMENTS A.
Test Cell The sample tank used for all the experiments was glass built and the total volume of this tank was 550 ml. A pair of spherical electrodes with a diameter of 13 mm have used for the experiments. The material of the electrodes is brass. The electrodes were attached to either side of the test cell wall from which they extended towards the middle of the cell. One of the electrodes are fixed length and the electrodes could be adjusted via a screw. The distance between the electrodes was kept constant at 10 mm for all the experiments.
I. INTRODUCTION
Power transformers are one of the key components in high voltage transmission and distribution systems. Their reliable operation is of paramount importance to energy generation sectors and users. Transformer failures have started to accelerate as more of them are reaching their designed lifetime. It has been previously reported by researcher [1] that almost one third of total transformer failures are caused by insulation failures. Understanding the failure mechanisms properly is the only way that will enable us to take proper measures to prevent failure.
B.
Sample Preparations Cellulose fibre dust was produced by rubbing a piece of new pressboard which normally used in high voltage transformer with different sizes of metal files. Different sizes of sieves were used to separate the fibres. Two different sizes of particles were used for these experiments, 63-150 μm for AC and DC biased AC, 150-250 μm for DC. The particles were separated by the fiber width rather than length. The contamination levels for each size of particles were 0.001%, 0.002%, 0.003%, 0.004%, 0.006%, 0.008%, 0.016% and 0.024% by weight. A digital measurement scale capable of measuring microgram was used to achieve maximum accuracy. For DC experiments only 0.001% to 0.004% contaminant were investigated whereas for AC all the above contamination levels used. In the case of DC biased AC, only 0.024% which was the highest contamination level was tested.
Liquid dielectric oils are used in many high voltage applications including majority of high voltage power transformers. They have two very important advantages of use, heat transfer and insulation but they also have a major disadvantage i.e. they are very easy to be contaminated [2, 3]. Transformer oil inside a transformer normally contacts with metal, iron core and pressboard insulation. Metal filings or cellulosic residual can be formed in transformer oil, especially for aged transformers with old pressboard insulation. Nonuniform fields are present within various area of transformer during operating condition. The contaminant particles tend to move towards high field regions due to dielectrophoresis (DEP) forces. Over a period of time these particles could form a bridge. The bridge may leading to partial discharges or insulation failure.
The sample tank was cleaned with soap solution in hot water then it was dried in hot air flow before starting a new test with a new size of particle. For repeating a test with same particle size, the sample tank was first rinsed with some clean oil then the test cell was rinsed thoroughly with cyclohexene.
Researchers have studied conduction current, partial discharge, resistivity [4 - 6] of bridging in oil. Effect of particle size [7] and mathematical modelling [8] of bridging have been previously reported by us. Three different experiments are performed to investigate the particle accumulation between two electrodes with different potentials under AC, DC and DC biased AC voltages. Three different levels of DC, AC and DC biased with AC voltages were investigated for these experiments. The effects of different level of contaminations ranging from 0.001 to 0.024% by weight were also accomplished. Optical images of pressboard particle concentration and conduction current were
A new experiment was always started with adding 300 ml of transformer mineral oil into the test cell which was enough to submerge the electrodes completely in oil. The lowest contamination level of pressboard fiber was then added to the oil. The test cell was covered with cling film to protect from dust and moisture. The sample tank was covered during whole experimental period apart from adding the next level of contaminants. The sample tank was stirred prior to every test
978-1-4799-2597-1/13/$31.00 ©2013 IEEE 943
on a magnetic stirrer for 2 minutes to disperse the particles evenly.
entire test to record the images of bridging. All tests were conducted at ambient room temperature. The temperature of the aluminum box was monitored during DC test. The temperature was within 3 degrees difference for all the DC experiments. All the tests were conducted three times for each voltage level to observe the repeatability of the obtained results.
C.
Experimental Setup The sample tank was positioned under a stereo microscope that had a digital camera mounted on the top to record the optical images of the particle accumulation. For experiments with DC electric field, the microscope along with the test cell was placed inside an aluminum box which acts as a Faraday Cage as the conduction current was too noisy. As for AC and DC biased AC test the aluminium box was not used. A desktop computer was used to control the digital camera and also store data from camera, Keithley Picoammeter 6485 (DC) and multimeter 2001 (AC) for conduction current measurement. A block diagram of complete experimental setup for DC biased AC test is shown in Fig. 1. A detail description of DC experiments can be found in [8].
III. RESULTS AND DISCUSSION A.
Bridge formation under the influence of DC The pressboard fibers started moving back and forward as soon as 2kV supply was turned on. They were charged by touching the positive electrode and discharging to the ground electrode. This process went on for a while then eventually the particles started to attach on the electrodes surface. The bridging starts with a longer fiber attached to the electrodes and elongate towards other electrode. Then small particles started attaching themselves to the long fiber. The bridge grows in this fashion. A thin complete bridge formed at 2kV after 180s and it continued to grow thicker until 600s as shown in Fig. 2.
This experimental setup for DC biased AC test consisted of a signal generator, high voltage amplifier, sample tank/test cell, microscope, digital camera and computer. A signal generator was used to produce the sinusoidal voltage of 50 Hz with a DC offset. This signal was amplified with the high voltage amplifier from Trek. The ratio of the amplification was 2000:1. The amplified signal was connected to the electrode. The high voltage amplifier also had an option to monitor the output voltage which was connected to an oscilloscope. The other electrode was connected to the ground. The conduction current was not measured for this experiment.
2kV DC at 120s
2kV DC at 180s
2kV DC at 600s
7.5kV DC at 10s
7.5kV DC at 30s
7.5kV DC at 300s
15kV DC at 5s
15kV DC at 15s
15kV DC at 60s
Fig. 2 Optical microscopic image for bridging under influence of DC electric field with 0.003% concentration of 150-250 μm cellulose particles
After applying 7.5 kV, the cellulose particles started moving more quickly. A thin complete bridge was observed within 10s of applied voltage. The bridge continued to grow thicker until 300s (Fig. 2) after that there was no significant change observed.
Fig. 1 Experimental setup for DC biased AC test
D.
Experiments under different electric fields One of the spherical electrodes was attached to the high voltage source. The other electrode was connected to the ground via a Keithley picoammeter 6485 (DC) and Keithley multimeter 2001 (AC) to measure the conduction current through the gap. The conduction current were not measured for DC biased AC test so the electrode was directly attached to ground. Three different voltage levels were investigated for each category of electric field, such as DC (2kV, 7.5kV and 15kV), AC (10kV, 15kV and 20kV peak-peak) and DC biased AC (1kV, 3kV and 6kV DC offset with 10kV, 15kV and 20kV AC peak-peak). Each experiments were carried out until a complete bridge was created between the electrodes or maximum of 25 minutes where there was no bridge. Optical microscopic images were taken in a regular interval during the
For 15kV, the contaminant movements were intensified rapidly. Within 5s, a thin complete bridge was formed and the bridge were thickened up to 60s. The bridge continued to grow thicker until 600s. The conduction currents under the influence of three different voltage levels are shown in Fig 3. Initially there was a high polarization current observer for some experiments then a slow increase observed until the current reaches saturation level. The saturation current reached when the bridge were roughly reached its maximum thickness. The conduction currents for all the contamination levels had a similar trend. The conduction currents increased a great deal with the
944
increment of particle contamination as the bridge also thickened with it.
As time elapsed, more particles were attached to both electrodes. Several contamination levels i.e. 0.001%, 0.002, 0.003, 0.006%, 0.008%, 0.016% and 0.024% were tested under influence of AC electric field. The particles accumulations increased to the electrodes surface increased with increament of contamination level. But a complete bridge between the two electrodes were never formed as for DC tests (Fig. 5). These results are contrary to the previous researcher reported [3] that bridging were observed with spherical and wrapped electrodes in moistened and pressboard fiber contaminated transformer oil under AC electric field of 6kV and 8 kV for spherical and formed wire electrodes respectively. Conduction currents were also recorded during the AC experiments but for the limited spaces it was not possible to show in this paper. The currents were in micro amp region which was significantly higher than DC. There was not significant change observed for a particular applied voltage with different concentration levels.
Fig. 3 Conduction currents under influence of different DC electric fields with 0.003% concentration
The images for particle accumulation were analysed using pixel counting technique. The very first image of a particular experiments was taken as background and it was subtracted from all the later images. The result were obtained as a grey image with a specified tolerance level which was fixed for all the analysis to obtain consistent result. Then resultant image were converted into black and white. After that all the black pixels were counted. The quantity of pixel count does not correspond to the particles number but the increment of the pixels shows the particles accumulation between the electrodes. The plots on Fig. 4 in below shows the pixel counts for optical microscopic image taken during the test. The graph clearly indicates the increments of particles with time. There is a close relation between the conduction current increments with particle accumulation observed by comparing the pixel count graph with conduction current. The plots has some glitches due to the LED light of the microscope. After a while the light became brighter which introduced an error to the pixel count and some of the significant errors removed manually from the data.
10kV AC at 1min
10kV AC at 10min
10kV AC at 25min
15kV AC at 1min
15kV AC at 10min
15kV AC at 25min
20kV AC at 1min
20kV AC at 10min
20kV AC at 25min
Fig. 5 Optical microscopic image for bridging under influence of AC electric field with 0.024% concentration of 63-150 150-250 μm cellulose particles
Fig 6 represents the pixel counts graph under influence of different AC electric field. The particle accumulation for 10kV was not significant because of weaker dielectrophoresis force. Almost similar amount of particles accumulated for 15 kV and 20kV AC electric field.
Fig. 4 Increment of pixels in microscopic images under influence of DC electric field with 0.003% concentration
B.
Bridge formation under the influence of AC The pressboard particles started moving slowly after the 10kV AC supply turned on. They fibers were drawing near to the high electric field region and attaching themselves to both electrodes surface slowly because of dielectrophoretic force.
Fig. 6 Increment of pixels in microscopic images under influence of AC electric field with 0.024% concentration
945
Fig 8 shows the graphs for the pixels counts under the influence of three different levels of AC electric field biased with 3kV DC. The pixels for 10kV AC did not increased much which shows similarity with the images. The particles increased after a while linearly for 15kV. The number of pixels for 20kV also showed increment quicker than the 15kV but the slope of the graph steeper than before.
C.
Bridge formation under the influence of DC biased AC There were many branches of bridge were formed within 30s switch on 3kV DC supply. The bridge was forming over a period of 15 min but it was a shallow bridge with lots of different branches which is on the first row of Fig 4. When the 3kV DC was combined with 10kV AC, the bridge was very strongly bonded. For 15kV and 20kV AC, the bond became stronger and more particles joined the bridge than 10kV. The particles which were on the side of the electrodes were going to discharge on the other side and eventually attracted to the middle of the bridge. That is how the bond of the bridge became stronger for the higher electric field.
3kV DC at 1min
3kV DC & 10kV AC at 1min
3kV DC at 5min
IV. CONCLUSION Several conclusions can be drawn from the above results. The particles accumulates towards the electrodes from all the DC, AC and DC biased AC electric field. As the voltage increased, the rate of bridge formation is intensified along with an associated current increase for DC electric field. The rate of particle accumulation for AC was slower at all voltage levels than DC. But the complete bridge between the electrodes only form when a steady DC voltage is applied. It is clear based on current results that charge transportation takes place through the cellulose particles under DC electric field. Although bulk conductivity of the pressboard is several hundred times less than the transformer mineral oil, the surface conductivity may be different to the bulk and which make the pressboard fibre to conduct under DC electric field. The properties of pressboard might have changed when they are in the state of fibre. Further investigations required on surface conductivity of pressboard which will reveal more clear indication of charge transfer phenomenon.
3kV DC at 15min
3kV DC & 10kV AC at 10min 3kV DC & 10kV AC at 25min
ACKNOWLEDGMENT The authors acknowledge the project financial support received from IET Power Academy and the National Grid.
3kV DC & 15kV AC at 1min 3kV DC & 15kV AC at 10min 3kV DC & 15kV AC at 25min
REFERENCES [1] [2]
3kV DC & 20kV AC at 1min 3kV DC & 20kV AC at 10min 3kV DC & 20kV AC at 25min
Fig. 7 Optical microscopic image for bridging under influence of DC biased AC electric field with 0.024% concentration of cellulose particles
[3]
There are three different DC offset levels investigated i.e. 1kV, 3kV and 6kV. These three levels of DC voltage showed that as the voltage increased, the thickness of the bridge also increases.
[4]
[5]
[6]
[7]
[8]
Fig. 8 Increment of pixels in microscopic images under influence of DC biased AC electric field with 0.024% concentration
946
V. V. Sokolov, “Experience with the refurbishment and life extension of large power transformers”. Minutes of the sixty-first annual conference of doble clients, Sec. 6-4, 1994. T. O. Rouse, “Mineral insulating oil in transformers.” Electrical Insulation Magazine”. IEEE 14(3): 6-16, 1998. M.G. Danikas, “Breakdown of transformer oil”. Electrical Insulation Magazine, IEEE. 6(5): p. 27-34, 1990. G. Chen and M. H. Zuber, “Pre-breakdown characteristics of contaminated power transformer oil”. in Electrical Insulation and Dielectric Phenomena, 2007. CEIDP 2007. Annual Report - Conference on, 2007, pp. 659-662. K. W. H. Moranda, H. M. Grzesiak, “Dynamics of bridge creating in contaminated oil at AC voltage and analysis of accompanying partial discharges”. XIII International Symposium on High Voltage Engineering Netherlands, 2003. J. G. M. Ossowski, K. W. H. Moronda and H. M. Grzesiak, “Oil resistance at different phases of bridge mechanism development at direct voltage”. XIII International Symposium on High Voltage Engineering Netherlands, 2003. S. Mahmud, I. O. Golosnoy, G. Chen, G. Wilson, and P. Jarman, “Numerical simulations of bridging phenomena in contaminated transformer oil”. Proceedings of 2012 IEEE Conference on Electrical Insulation and Dielectric Phenomena, Montreal, Canada, 14 - 17 Oct 2012.4pp, 383-386. S. Mahmud, G. Chen, I. O. Golosnoy, G. Wilson, and P. Jarman, “Bridging phenomenon in contaminated transformer oil”. Proceedings of 2012 International Conference on Condition Monitoring and Diagnosis, Piscataway, US, Institute of Electrical and Electronics Engineers, 4pp, 180-183.
