Mechanisms of Electrical Ageing of Oil- impregnated Paper due to Partial Discharges. MOHAMAD GHAFFARIAN NIASAR. Doctoral thesis in Electrical Systems.
KTH Electrical Engineering
Mechanisms of Electrical Ageing of Oilimpregnated Paper due to Partial Discharges
MOHAMAD GHAFFARIAN NIASAR
Doctoral thesis in Electrical Systems Stockholm, Sweden 2015
Division of Electromagnetic Engineering KTH School of Electrical Engineering SE– 100 44 Stockholm, Sweden
TRITA-EE 2015:006 ISSN 1653-5146 ISBN 978-91-7595-443-1 Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen fredagen den 27 februari 2015 klockan 14.00 i H1, Teknikringen 33, 1 tr, Kungl Tekniska högskolan, Stockholm. © Mohamad Ghaffarian Niasar, February 2015 Tryck: Universitetsservice US AB ii
Abstract In this thesis, partial discharge (PD) phenomenon in oil-impregnated paper (OIP) is investigated under accelerated electrical stress. The thesis is mainly focused on the characteristic of PD activity and the influence it has on the insulation properties of OIP. PD source was created by introducing an air filled cavity embedded between layers of OIP. PD activity is investigated from the initiation up to final puncture breakdown of the OIP. The time-evolution of number, maximum magnitude and average magnitude of PD is investigated for cavities with different diameter and height. It was found that time to breakdown is shorter if the cavity diameter is larger and cavities with higher depth produce larger PDs. Comparison between PD activity in three cases, i.e. unaged OIP, thermally aged OIP and OIP samples with higher moisture content is performed. In general, it is found that for all cases the number and the maximum magnitude of PD follows a similar trend versus ageing time. During the very beginning of the experiment large discharges occur and they disappear after a short ageing time. Number and maximum magnitude of PD increase with time until reaching a peak value. Finally both parameters decrease with time and puncture breakdown occurs in the sample. Even though PD activity in thermally aged OIP is higher compared to the unaged OIP samples, the time to breakdown for new and thermally aged OIP samples is similar while it is shorter for OIP samples with higher moisture content. The influence of different parameters and in particular the influence of PD on dielectric frequency response (DFR) of OIP is investigated. It is demonstrated that PD activity permanently change the DFR of OIP and the change mostly occurs in the frequency range between 1 mHz – 1 Hz. The change of DFR due to PD activity is very similar to the change occurring due to thermal ageing and at lower frequencies it is also similar to the change caused by higher moisture content. The change is also similar to the case when the oil layer on top of OIP has higher conductivity. Breakdown strength of OIP samples is measured before and after ageing with PDs. It is found that the breakdown strength of OIP samples decreases by around 40% after the sample is exposed to accelerated electrical ageing. Furthermore a thermal model was developed to investigate the possible transition of breakdown mechanism from erosion to thermal breakdown in OIP dielectrics. It was found that PD activity can lower the thermal breakdown voltage of OIP up to four times. Keywords: partial discharges, oil-impregnated paper, oil, transformer insulation, dielectric frequency response, breakdown strength, ion mobility, thermal ageing, electrical ageing, moisture in paper, life time, thermal breakdown. iii
Sammanfattning I denna avhandling studeras accelererad elektrisk åldring av oljeimpregnerat papper (OIP) under inverkan av partiella urladdningar (PD). Avhandlingen är främst inriktad på karaktäristiska egenskaper hos PD aktiviteten och inverkan av PD på åldringen av OIP. PD källor skapades genom att luftfyllda hålrum inbäddades mellan lager av OIP. PD-aktiviteten undersöktes från initiering till slutgiltigt genomslag. Tidsutvecklingen av antal PD per spänningscykel, maximal storlek och genomsnittliga storleken på PD studerades för kaviteter med olika diameter och höjd. Stora kaviteter uppvisade en kortare tid till genomslag. Högre kaviteter medförde större urladdningar. En jämförelse gjordes för PD aktivitet i tre fall: oåldrad OIP, termiskt åldrad OIP och OIP prover med 5,5% fukthalt. I allmänhet visar resultaten att antalet PD per period och den maximala storleken av PD följer en liknande trend i förhållande till åldringstiden. Under början av åldringen är urladdningarna stora men de minskar snabbt efter en kortare tids åldrande. Antal per period och maximal PD storlek ökar med tiden tills de når ett maximum. Därefter minskar båda parametrarna kontinuerligt med tiden tills genomslag sker. Även om PD aktivitet i prover med kaviteter av termiskt åldrad OIP är högre jämfört med de icke åldrade OIP proverna så är tiden till genomslag under PD lika för initialt oåldrade prover och termiskt åldrade prover, medan den är kortare för OIP prover med högre fukthalt. Påverkan av olika parametrar och i synnerhet påverkan av PD på dielektriska frekvensresponsen (DFR) av OIP har studerats. Resultaten visar att PD aktivitet i en kavitet ledet till permanenta förändringar av de dielektriska egenskaperna av den omkringliggande OIP isolationen och är vid rumstemperatur tydligast inom frekvensområdet mellan 1 mHz - 1 Hz. Förändringen av DFR grund av PD aktivitet är mycket lik den förändring som inträffar på grund av termisk åldring och vid lägre frekvenser är den också lik den förändring som orsakas av högre fukthalt. Beteendet liknar också den dielektriska respons som fås av ett oljeskikt med hög konduktivitet i serie med OIP. Genomslagshållfastheten av OIP prover mätta före och efter åldring med PD visar att hållfastheten minskar med cirka 40% av den accelererade elektriska åldringen. En termisk modell utvecklades för att undersöka den möjliga övergången av nedbrytningsmekanism från erosion till termisk nedbrytning i OIP. Resultaten visar att PD-aktivitet kan sänka den termiska genombrottsspänningen hos OIP med upp till fyra gånger. Ämnesord: partiella urladdningar, oljeimpregnerat papper, transformator isolation, dielektrisk frekvensrespons, genomslags, jon mobilitet, termisk åldring, elektrisk åldring, fukt i isolation, livslängd, termiskt genomslag iv
Acknowledgements I am very grateful to have had the opportunity to do this PhD thesis at KTH. I would like to thank my supervisor, Assoc. Prof. Hans Edin, who taught me how to become an independent researcher. His professional behavior and excellent guidance in this project makes all difficulties easy to deal with. Thanks to Dr. Nathaniel Taylor who is currently a researcher in the Electromagnetic Engineering Department in KTH. He was always ready to help and his assistance with laboratory equipment and discussion about practical and theoretical problems was excellent. Dr. Taylor also did proof-reading of this thesis with many valuable comments. I also like to thank Prof. Rajeev Thottappillil, head of the department, for all his efforts to make the department a nice place to work. I have spent six months of my PhD period at Institute of Power Transmission and High Voltage Technology (IEH) of University of Stuttgart. This period was a mandatory part of EIT/KICInnoEnergy PhD school and turned out very fruitful to me. I want to thank Prof. Stefan Tenbohlen for accepting me in his research group. I also want to thanks Dr. Urich Schärli for arranging the administrative issues related to this mobility in Germany. I am very grateful to my friends in the high-voltage research group, Dr. Nadja Jäverberg, Dr. Respicius Clemence, Tech. Lic. Xiaolei Wang, Tech. Lic. Roya Nikjoo, Patrick Janus and Håkan Westerlund and all other colleagues in Teknikringen 33 for creating a very friendly atmosphere to work in. Thanks to my Persian friends, Seyed Ali Mousavi, Babak Kairi and others, for helping me to enjoy the life in Sweden more efficiently. Also I would like to thank our financial administrator Ms. Carin Norberg, computer system administrator Mr. Peter Lönn, and workshop responsible Mr. Jesper Freiberg. Furthermore, I acknowledge ABB for supplying paper and pressboard, ABB Power Transformers in Ludvika for performing gas in oil analysis of my oil samples, NynäsAB for supplying transformer oil and Vattenfall/Forsmark nuclear power station for supplying an aged transformer bushing. The first three years of the project were funded by the Swedish center of Excellence in Electric Power Engineering – EKC and the last two years was funded by the Swedish Center for Smart Grids and Energy Storage – SweGRIDS, which is greatly acknowledged. I am grateful that the project became a part of the innovation project v
CIPOWER within EIT/KICInnoEnergy. KICInnoEnergy PhD School provides us with number of useful courses and other supports which is greatly appreciated. During these 23 years of being student, I have had many teachers who worked hard to educate me. I might not remember their entire names, but I am always aware of the great influence they have had on my life. I say thanks to all teachers. Being far from my family for nearly 5 years, I can better understand how valuable they are. I believe my parents did a magnificent job on educating their kids even though they had little resources available. Hereby I say thanks to my father and my mother for their parental advice. During my normal life and educational career my sisters and brothers have always been supportive to me by any means. Thanks and I hope I can compensate it sometime in the future. Finally, thanks to my best friend, Mohamad, who is always talking to me and giving me ideas to better approach the challenges of my life.
Mohamad Ghaffarian Niasar Stockholm, February 2015
vi
List of publications This work is based on following papers: I.
M. Ghaffarian Niasar, R. Clemence Kiiza, N. Taylor, P. Janus, X. Wang, Hans Edin, “Partial Discharges in a Cavity Embedded in Oil- Impregnated Paper: Effect of Electrical and Thermal Aging”, accepted for publication in IEEE Transactions on Dielectrics and Electrical Insulation.
II.
M. Ghaffarian Niasar, N. Taylor, R. Clemence Kiiza, Hans Edin, “Dielectric Frequency Response of Oil-impregnated paper: the Effect of Partial Discharges Compared to other Influences”, submitted to IEEE Transactions on Dielectrics and Electrical Insulation.
III.
R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin, “Change in Partial Discharge Activity as Related to Degradation Level in Oilimpregnated Paper Insulation”, IEEE Transactions on Dielectrics and Electrical Insulation Vol. 21, No. 3; June 2014.
IV.
R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin, “The effect of PD by-products on the dielectric frequency response of oilimpregnated paper insulation comprising of a small cavity”, submitted to IEEE Transactions on Dielectrics and Electrical Insulation.
V.
M. Ghaffarian Niasar, R. Clemence Kiiza, X. Wang, R. Nikjoo, Hans Edin, “Aging of Oil Impregnated Paper Due to PD Activity”, 18th International Symposium on High Voltage Engineering (ISH 2013), Seoul, South Korea.
VI.
M. Ghaffarian Niasar, R. Clemence Kiiza, N. Taylor, Hans Edin, “Effect of Partial Discharges on Thermal Breakdown of Oil Impregnated Paper”, Submitted to IEEJ Transactions on Fundamentals and Materials.
Licentiate thesis: M. Ghaffarian Niasar, Partial Discharge Signatures of Defects in Insulation Systems Consisting of Oil and Oil-impregnated Paper, Lic. Thesis, KTH, 2012, [Online]. Available: http://www.diva-portal.org/smash/get/diva2:572145/FULLTEXT01.pdf vii
Papers included in Licentiate thesis:
VII.
M. Ghaffarian Niasar, H. Edin, “Corona in Oil as a Function of Geometry, Temperature and Humidity” Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), October 2010, West Lafayette, USA.
VIII.
M. Ghaffarian Niasar, H. Edin, “Partial Discharge Due to Bubbles in Oil” Nordic Insulation Symposium, June 2011, Tampere, Finland.
IX.
M. Ghaffarian Niasar, H. Edin, X. Wang and R. Clemence, “Partial Discharge Characteristics Due to Air and Water Vapor Bubbles in Oil” 17th International Symposium on High Voltage Engineering, August 22nd-26th 2011, Hannover, Germany.
X.
M. Ghaffarian Niasar, R. Clemence, X. Wang, R. Nikjoo, H. Edin, “Effect of Temperature on Surface Discharge in Oil” IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), October 2012, Montreal, Canada.
The author has also contributed to the following papers during his PhD period. XI.
X. Wang a, R. Clemence Kiiza, M. Ghaffarian Niasar, H. Edin, “Surface charge dynamics studied by the temporal evolution of the corona charging current”, Submitted to Journal of Electrostatics, Elsevier, 2014.
XII.
M. Ghaffarian Niasar, R. Clemence Kiiza, Hans Edin, “Oil Aging Due to Partial Discharge Activity”, Nordic Insulation Symposium- Nord-IS 13 Trondheim, Norway, June 9-12, 2013.
XIII.
R. Clemence Kiiza, M. Ghaffarian Niasar, R. Nikjoo, X. Wang and H. Edin, “Comparison of Phase Resolved Partial Discharge Patterns in Small Test Samples, Bushing Specimen and Aged Transformer Bushing” IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), October 2012, Montreal, Canada.
XIV.
X. Wang, M. Ghaffarian Niasar, R. Clemence, H. Edin, “Partial Discharge Analysis in a Needle-plane gap with Repetitive Step Voltage” IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), October 2012, Montreal, Canada. viii
XV.
R. Clemence Kiiza, R. Nikjoo, M. Ghaffarian Niasar, X. Wang and H. Edin, “Effect of High Voltage Impulses on Partial Discharge Activity in a Cavity Embedded in Paper Insulation” IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), October 2012, Montreal, Canada.
XVI.
R. Nikjoo, N.Taylor, M. Ghaffarian Niasar, H. Edin, “Dielectric Response Measurement of Power Transformer Bushing by utilizing High Voltage Transients” IEEE Conference on Electrical and Dielectric Phenomena (CEIDP), October 2012, Montreal, Canada. X. Wang, R.C. Kiiza, M. Ghffarian Niasar, R. Nikjoo, H. Edin, “Effect of dielectric material on decay of surface charge deposited by corona discharge”, ISH 2013, Seoul, South Korea.
XVII.
XVIII.
Demetres Evagorou, Patrick Janus, M. Ghaffarian Niasar, Hans Edin, “Identification of PD sources using laboratory measurements”, ISH 2013, Seoul, South Korea.
XIX.
R. Nikjoo, N. Taylor, R.C. Kiiza, M. Ghffarian Niasar, X. Wang, H. Edin, “Insulation condition diagnostics of oil impregnated paper by utilizing power system transients”, ISH 2013, Seoul, South Korea.
ix
x
Contents 1
2
Introduction .......................................................................................................... 1 1.1
Background ................................................................................................... 1
1.2
Aim ............................................................................................................... 3
1.3
Project aim .................................................................................................... 4
1.4
Thesis outline ................................................................................................ 4
1.5
Author’s contributions .................................................................................. 5
Literature overview and theoretical background .............................................. 7 2.1
Non electrical ageing mechanisms of paper insulation ................................. 7
2.2
Electrical ageing and partial discharge measurement ................................... 9
2.3
Dielectric Spectroscopy .............................................................................. 11
2.3.1
Theory .................................................................................................... 11
2.3.2
Diagnostic tool........................................................................................ 14
2.4 3
4
Breakdown mechanism in solid insulations ................................................ 17
Measurement systems and sample preparation ............................................... 19 3.1
PD measurement system ............................................................................. 19
3.2
Dielectric frequency response measurement ............................................... 23
3.3
Polarization and depolarization current measurement ................................ 25
3.4
Sample preparation ..................................................................................... 26
Evolution of PDs with time for a cavity embedded between layers of OIP ... 27 4.1
Electrical aging of OIP through exposure to PD ......................................... 27
4.2
Limitations of the measurement .................................................................. 31
4.3
Variation of time to breakdown for similar samples ................................... 33
4.4
Effect of cavity geometry on PD parameters and time to breakdown ......... 34
4.4.1
Cavity diameter ...................................................................................... 34
4.4.2
Cavity height .......................................................................................... 35
4.5
Effect of thermal ageing and higher moisture content on PD parameters ... 37
4.6
Time to breakdown ..................................................................................... 38 xi
5
6
Dielectric frequency response of oil-impregnated paper ................................ 39 5.1
Effect of OIP samples and measurement conditions ................................... 39
5.2
Effect of voltage and measurement of ion mobility .................................... 42
5.3
Effect of moisture content of the paper ....................................................... 49
5.4
Effect of thermal ageing .............................................................................. 53
5.5
Effect of temperature .................................................................................. 56
5.6
Effect of Oil contamination......................................................................... 57
5.7
Effect of other parameters ........................................................................... 58
5.8
Effect of electrodes ..................................................................................... 59
5.9
Effect of PD ................................................................................................ 60
5.10
Effect of PD on each layer .......................................................................... 62
5.11
Contamination/memory effect of the electrodes after exposure to PD ....... 63
5.12
Effect of cavity size on DFR ....................................................................... 65
5.13
Effect of electrical discharge in oil on oil properties .................................. 66
5.14
Effect of PD on OIP with high level of moisture ........................................ 68
5.15
Effect of PD on thermally aged paper ......................................................... 69
5.16
Effect of gas production due to PD ............................................................. 69
5.16.1
Experiment with brass electrodes performed under oil ...................... 70
5.16.2
Experiment with DFR test cell performed under oil .......................... 70
Effect of PD on breakdown strength of OIP and oil ........................................ 73 6.1
AC and Impulse breakdown strength of OIP .............................................. 73
6.1.1
AC breakdown strength of oil-impregnated paper.................................. 73
6.1.2
Impulse breakdown strength of oil-impregnated paper .......................... 75
6.1.3 AC breakdown strength of oil-impregnated paper with a cavity in between layers of paper ........................................................................................ 75 6.1.4 Impulse breakdown strength of oil-impregnated paper with a cavity in between layers of paper, before and after ageing with PD ................................... 76 6.1.5 6.2 7
Effect of gas layers on impulse breakdown strength .............................. 77 Breakdown strength of oil ........................................................................... 78
Thermal modeling .............................................................................................. 81 xii
7.1
FEM modeling procedure for thermal modeling of dielectrics ................... 82
7.1.1
Step 1: Calculation of the heat source density inside the dielectric. ....... 82
7.1.2
Step 2- Calculation of activation energy for the insulation .................... 85
7.1.3
Step 3- Selecting the correct boundary conditions ................................. 87
7.1.4 Step 4- Solving of the coupled system of heat transfer and electric potential equations ................................................................................................ 88 7.2
Simulation results........................................................................................ 89
7.2.1
Thermal breakdown voltage of unaged OIP samples ............................. 89
7.2.2
Thermal breakdown voltage of OIP samples after ageing with PD ........ 89
8
Summary of the papers ...................................................................................... 93
9
Conclusions and future work............................................................................. 97
Bibliography ............................................................................................................... 99 Appendix A ............................................................................................................... 107 Appendix B................................................................................................................ 111
xiii
xiv
1 Introduction 1.1 Background Transformers are one of the main components of any distribution and transmission electric power system. They are large and expensive and therefore replacing old units with new ones with the purpose of increasing the reliability of the power system might not be economically justified. It is therefore strongly desired that old units stay in operation for longer time. However, the risk of failure increases with the age of the transformer and therefore keeping the old transformers in service is associated with a high risk. The cost of unexpected failure of a power transformer might be up to several times the initial cost of that transformer [1]. There is not only replacement cost but also possible cost related to cleaning, deterioration of power quality and loss of revenue due to power shut down. Therefore, in order to keep the units in service while limiting the risk of failure, diagnostic measurements can be used. Diagnostic measurements can be performed on-line or off-line during the maintenance period. Available diagnostic methods can address electrical, thermal and mechanical conditions of a transformer. Partial discharge measurements can detect an incipient fault in the insulation system. Thermal assessments can be used to analyze the remaining lifetime of the insulation system. Mechanical methods such as frequency response analyses (FRA) can be used to identify deformation in windings, shorted turns etc. [2]. There is no diagnostic method that can certainly predict the time to failure of the insulation system. However, the probability of detecting an important problem can be increased by combining the predictions and results of different diagnostic methods. Monitoring a change in the transformer can be performed through a number of diagnostic methods such as: dissolved gas in oil analysis (DGA), degree of polymerization (DP), infrared thermography analysis, FRA, partial discharge (PD) measurements, dielectric spectroscopy (DS) and polarization/depolarization current analysis (PDC), loss factor measurement at power frequency, winding resistance etc. [2]. A study of transformer failures carried out on transformers rated at 25 MVA or above for the period 1997 till 2001 shows that the insulation breakdown is the leading cause of transformer failure [3]. Table 1.1 shows the number of failures and the cost of repair reported by that study. Insulation failure in that study includes
2
insulation degradation, defective insulation and short circuits, but excludes line surges and line faults. Another study [4], reports that the responsible failure mechanism is related to tap changer (41%), winding (19%), tank leakage (13%), bushing (12%), core (3%) and other (12%). Table 1.1: Cause of failures [3] Cause of failure
Number
Insulation Failure Design/Material/Workmanship Unknown Oil contamination Overloading Fire/Explosion Line Surge Improper Maintenance/Operation Flood Loose connection Lightning Moisture Total
24 22 15 4 5 3 4 5 2 6 3 1 94
Total Paid [M USD] 150 65 30 12 8.6 8 5 3.5 2.3 2.2 0.66 0.18 287.44
The majority of power transformers in service are oil filled. Oil and oil-impregnated paper (OIP) are the main insulation in the oil filled transformers and in the connected OIP bushings. Oil is the main element of the transformer cooling system, and it also provides good electrical insulation when combined with the cellulosebased insulation. Remarkable electrical and mechanical properties, availability and low cost, and experience of reliable and safe operation are the main reasons that OIP has been used in transformer and bushing manufacturing for nearly a century. Partial discharge measurement is a non-destructive diagnostic method that can detect the existence of external corona or internal insulation defects. It is performed as a routine test by transformer factories on newly built transformers before final delivery. A correlation of PD with insulation lifetime has been recognized for a long time, yet still there is not a general model that can accurately correlate the PD activity with the lifetime of insulation. PD deteriorates insulation material through mainly three mechanisms, i.e. the impact of high energy electrons or ions with the insulation, formation of chemical byproducts, and Ultra-Violet (UV) rays or soft Xrays [5].
3
In the literature, many studies [5-10] are performed on ageing of polymer material and machine insulation exposed to PD. For transformer insulation, the phenomenon of surface discharges on the interface of oil-and pressboard was investigated in a number of articles [11-19]. However, there has been less attention to ageing of transformer/bushing insulations due to internal discharges [20]. The ability to interpret partial discharge measurement results is important for assessment of the degree of ageing of the insulation. Understanding the behavior of PD parameters during ageing time, and the possible correlation of those parameters with aging degree and lifetime of the transformer insulation, is essential in providing a good diagnostic power. Therefore, in order to make PD measurement a more accurate diagnostic tool for transformers, it is necessary to perform such studies. Dielectric Frequency Response (DFR) is another common method that is used in power transformers for evaluation of the insulation condition. Parameters such as moisture content, temperature, thermal ageing, oil acidity etc. can affect the DFR measurements. It is therefore possible to evaluate the insulation ageing status or moisture content of the paper by using DFR together with suitable mathematical modeling. However, other parameters such as PD activity and electrode material can also affect the DFR measurement results. The effect of PD on DFR was first introduced by the high voltage group of KTH and is tackled in the PhD thesis [21] of one of the co-authors of the attached articles. In this thesis a more detailed study is made of the influence of PD on DFR. A thorough investigation about the effect of various parameters on DFR of OIP and oil is performed and compared with the effect of PD.
1.2 Aim The aim of this thesis is provide an evaluation of the mechanisms behind electrical ageing of oil-impregnated papers insulation caused by partial discharges and thermal ageing. A further aim is to investigate how the PD ageing is affected by preceded thermal ageing and the humidity level of the oil-impregnated paper. The main objective of the study was to study how the dielectric frequency response of the oil-impregnated system was affected by the PD in comparison to other influences.
4
1.3 Project aim The aim of this PhD work is to provide a better understanding of PD activity in OIP insulation, and the effects that it has on the insulation system. The focus therefore must be mainly on OIP and oil insulation. The PhD work was divided into two parts. The objective of the first part was to design the laboratory models that can simulate the real defects in a transformer, and to investigate the PD patterns and waveform corresponding to each defect. In the licentiate work, different laboratory models were built. The models were: corona in air, corona in oil, surface discharge in air, surface discharge in oil, cavity discharges, discharge due to floating metal object in oil, discharges due to floating particles in oil and discharges due to moving and stationary bubbles in oil. PD pattern and waveform from each PD source was investigated. The influence of temperature was investigated on corona in oil, floating metal object in oil, air bubble adjacent to pressboard and surface discharge in oil. The influence of moisture content was only investigated on corona in oil. The possibility of PD classification based on PD pulse waveform and the phase of occurrence must be evaluated. This part of the PhD work is already published in the licentiate thesis presented in December 2012 [22]. The objective of the second part of the thesis is to analyze the ageing phenomenon of oil and OIP due to exposure to PD. The study must reveal the behavior of PD parameters over time when the insulation is exposed to PD. The effect of PD on insulation parameters such as conductivity, dielectric loss and breakdown strength are investigated.
1.4 Thesis outline This thesis contains eight chapters and is based on six papers that are attached to the end of the thesis. The disposition of the thesis is given here. The background and motivation to the subject and the aim of the study are given in Chapter 1. The outline of the thesis and the author’s contribution are also provided in this Chapter 1. Furthermore, a list of published papers with author’s contribution is given in Chapter 1. A literature review of the previous research about insulation diagnostics is given in Chapter 2. A short summary of the thermal ageing process of OIP and a review of the previous research on erosion effects of PD are provided in Chapter 2. A literature review related to DFR and a brief overview of the breakdown mechanisms in solids are also given in Chapter 2.
5
In Chapter 3, experimental setups and measurement equipment used in this work are described and explained. Ageing of OIP through PD exposure is investigated in Chapter 4. The effects of cavity geometry, thermal ageing and moisture content of the OIP on PD parameters are investigated. Evolution of PD parameters (number, maximum magnitude and average magnitude of PD) from inception of PD until occurrence of final breakdown in the OIP sample is also presented in this chapter. DFR measurements of OIP and the effect of parameters such as thermal ageing, temperature, moisture content, voltage, thickness of the sample, PD and electrode material is presented in Chapter 5. In Chapter 6, measurement results of breakdown strength of new and thermally aged OIP under sinusoidal 50 Hz and Lightning Impulse (LI) voltage was investigated. The effect of PD on reduction of impulse breakdown strength of OIP samples is presented and discussed. Chapter 7 contains a FEM thermal model which was developed in COMSOL and investigates the possible transition of breakdown mechanism from erosion (caused by PDs) to thermal breakdown on OIP insulation. Reduction of thermal breakdown voltage of OIP samples due to PD exposure is investigated in this chapter. A summary of published/submitted papers (that are attached to this thesis) is presented in Chapter 8. General conclusions and suggestions for future work are given in Chapter 9.
1.5 Author’s contributions In this thesis the electrical ageing of OIP due to exposure to PD activity is investigated in detail. The influence of PD activity on dielectric loss and breakdown strength of OIP is investigated both experimentally and by means of FEM simulation. Generally it was found that dielectric loss of OIP is significantly increased due to PD activity and the impulse breakdown strength of OIP reduces after exposure to PD activity. The author is fully responsible for paper I, II, V and VI. The co-authors on these papers mainly contributed to the revision of the paper. The author together with Respicius Clemence Kiiza performed most of the experiments and data analysis in papers I and III. In paper IV, the author mainly participated in discussion of the simulation results.
6
2 Literature overview and theoretical background 2.1 Non electrical insulation
ageing
mechanisms
of
paper
The lifetime of a power transformer is expected to be more than 30 years. Prior to delivery to customer, the factory tests ensure that the transformer is working well. These tests are divided into three categories i.e. type tests, routine tests and special tests. Type tests are conducted on a transformer that is representative of other transformers, and consist of several tests including temperature rise, and lightning impulse tests. Routine tests have to be performed on each individual transformer. These tests include transformer winding resistance measurement, transformer ratio test, transformer vector group test, measurement of impedance voltage/short circuit impedance and load loss (short circuit test), induced AC voltage test, partial discharge measurement, measurement of no load loss and electric current (open circuit test), separate source AC withstand voltage test, measurement of insulation resistance, dielectric tests of transformer, tests on on-load tap-changer, and oil pressure test. For example according to IEC 600763 [23] standard the acceptance PD level is 300 pC and 500 pC at 130% and 150% of rated voltage for an oil filled transformer. Special tests are performed upon the request of the customer which may include dielectric tests, switching impulse (SI) voltage test, measurement of zero sequence impedance, measurement of the harmonics of the no-load current, measurement of the dissipation factor (tan ), anti-corrosion protection test, measurement of acoustic noise level, measurement of power taken by fans and oil pumps, tests on components/accessories such as Buchholz relay, pressure relief valve, temperature indicators etc. During the in-service lifetime of the transformer the insulation system is exposed to electrical, mechanical, thermal and ambient stresses. The ageing mechanism of oil and paper is a complex phenomenon that can be accelerated by the presence of oxygen, water and acids. The byproducts of paper ageing are water, furanic components, gases, acids, and methanol. Due to ageing of paper, fibers may break from the paper and be suspended in the oil. Alignment of this fiber between conductors can make a path for flashover between conductors. Water in paper
8
increases the dielectric losses which in extreme cases can increase the local temperature above the permissible limit. Paper with 4% moisture content may age up to 20 times faster than dry paper [24, 25]. It is reported [25] that the degradation rate of paper can be reduced 16 times if the amount of oxygen is reduced from 2000 ppm to 300 ppm. The effect of acids on paper ageing depends on dissociation rate of the acid. When acids dissociate they produce H+ which increases the depolymerization rate of the paper. Paper can degrade through oxidation, hydrolysis and pyrolysis. References [24] and [26] provide a thorough explanation of each mechanism. Common methods that are used to assess paper ageing are: DP, tensile strength, furanic compound analysis and DGA. It is suggested [27–29] that the main paper ageing mechanism in an operating transformer is acid catalyzed hydrolysis. Hydrogen peroxide is created from the interaction of metals cations such as Cu+ and Cu++ or Fe+ and Fe++ together with heat, oxygen and water. The hydroxyl (HO) radical, which is a byproduct of decomposition of hydrogen peroxide, will catalyze the oxidation of paper and oil. The mechanical strength of paper insulation is one of the most important parameters of the OIP insulation. During short circuits faults, high current passing through the transformer winding exerts a large force on the winding, which affects the paper insulation system. If the OIP is not mechanically strong, the short circuit force can destroy the insulation system and lead to transformer failure. Mechanical strength of the paper can be associated to DP. Unprocessed paper insulation has DP in the range of 1000-1400 which reduces to 900-1100 with moisture content of 0.5 percent after being exposed to drying and impregnation procedure. When the DP value of paper insulation reduces to 200 it is commonly accepted that the tensile strength of the paper is around 20% of the original paper and this is often considered the end of life for the paper. A lot of research has been performed to show the correlation between DP and thermal ageing [30, 31]. Depending on the moisture content of paper and the ageing temperature the expected life of a paper insulation system can vary from a few months to a few hundred years. The expected life of cellulose insulation as a function of DP value can be shown by equation 2.1[2]: Expected life (years) =
∙
(2.1)
and DP are DP values of the OIP at the end and at the In equation 2.1 DP beginning of the lifetime, E is the activation energy (around 111 kJ/mole) R is the is a gas constant (8.314 J/mole/°K). T is temperature in degree Kelvin and coefficient which depends to type of insulation and operation condition. is
9
affected by moisture, oxygen and acidity of the oil. A typical value of for Kraft paper in dry and clean condition is 2.0 0.5 10 . More detail about the factor can be found in [2]. For example reference [2] shows that for dry and clean kraft paper and an operation temperature of 70 °C the expected life can be around 200 years while if the paper has 3-4% moisture the life expectancy at 70 °C is less than 20 years. What is mentioned in the above paragraphs is a summary of non-electrical ageing mechanisms of paper insulation. However, the main focus of this thesis is on electrical ageing of oil and OIP. Electrical ageing is commonly caused by PD activity and if it continues for enough time, depending on the type of defect, it may lead to insulation puncture or continuous surface flashover.
2.2 Electrical ageing measurement
and
partial
discharge
Over the years, much work has been done on PD measurement, which can generally be divided into four groups: detection, localization, classification and ageing. PD detection is based on the exchange of energy due to PD occurrence. It can be done electrically by means of electrical pulse current, electromagnetic radiation, dielectric loss or through non-electrical methods such as light detection, sound detection, increased gas pressure or chemical reactions. Identifying the location where PD comes from is done through PD localization. In large equipment such as power transformers, cables, gas insulated lines (GIL) and gas insulated substations (GIS), localization of PD is very important as it is very difficult to fix the problem without knowing where it is. Acoustic PD localization is a technic that is used for PD localization in power transformers [32–33] and [78]. It is usually important to identify and localize the source of PD. Some PD sources can be dangerous and deteriorate the insulation fast, while others can be active for many years without making too much trouble. In order to do an accurate classification it is important to know features of the PD signal corresponding to each defect. This is normally done through building laboratory setups to simulate most common possible situations in real equipment. The characteristics of PD signal due to different sources must be extracted, compared and classified first. The most suitable features are selected for the classification purpose. Using those features it is possible, with a good probability, to predict the source of the PD. These features can be phase of occurrence, shape of the PD pulse signal, PD
10
repetition rate, PD magnitude, symmetric or unsymmetrical phase resolved pattern etc. [22] and [34, 35]. One important question that a diagnostic method is supposed to answer is whether an insulation failure is imminent or not. Works that focused on PD ageing are targeting this question. As PD activity continues, it deteriorates the insulation. However, the rate of deterioration can be very different depending on the insulation material, intensity of PD, location of PD, influence of other parameters, etc. It is very important to access the information that correlates insulation ageing to PD activity. By assuming a correlation between ageing and lifetime, then there must be a correlation between PD activity and lifetime of insulation. The main mechanisms that PD deteriorates insulation are believed to be ion and electron bombardment and chemical reactions together with chemical byproducts that cause insulation degradation. The main ageing process of polymeric material induced by PD inside a cavity includes few stages as explained below [10].
The surface conductivity of the cavity increases due to byproducts produced by PD. For example, the increase of surface conductivity in epoxy is reported to be more than six orders of magnitude after exposure to PD [36]. Increase of surface roughness occurs due to ion and electron bombardments and chemical reactions. After long time of ageing, solid byproducts form on the surface of the insulation. Field enhancement occurs on the tips of crystals which later lead to formation of the first branch of electrical treeing. Growth of electrical tree leads to final breakdown.
The surface erosion may be caused by ion bombardment, UV light or X-rays [5]. Another possible explanation for increase of surface roughness due to PD exposure is that some electrons deposited from the previous discharges may be trapped slightly below the surface of the insulation and therefore ions produced by the later discharges cannot neutralize them. The very short distance between positive and negative charges causes very high local electrical field which may exceed the intrinsic breakdown of the solid insulation and leads to a local breakdown [5]. This process continues and leads to surface erosion. Byproducts produced by PD may change the gas properties and therefore change the discharge mechanism. It is reported [10, 20] that some PD byproducts in oxygen contain gas can reduce the PD magnitudes. Change in cavity surface conductivity and surface roughness may change the rate of ion injection and charge relaxation
11
which overall can affect the PD behavior. Depending to the type of dielectric, the mechanism that leads PD to breakdown differs. For example for the case of polyethylene, thermal degradation is dominant, while for mica insulation ion bombardment is believed to be dominant [5]. Glass and mica are very PD resistant while PVC and polyethylene are less resistant and rubber insulation can be damaged very fast under PD exposure. The deterioration rate approximately increases with the increase of number of discharges and therefore it is proportional to frequency of the applied voltage [5]. The lifetime is inversely proportional to the frequency of the applied voltage (at least up to certain frequencies such that high discharge intensity would not lead to thermal breakdown). In the case of DC stress the number of PD is much lower compared to the AC stress. That is why dielectric lifetime under DC stress can be longer than AC stress, even if though the amplitude of DC stress would be higher. The deterioration rate is also affected by the intensity of discharges which means at higher electric field strength the lifetime would be shorter. Higher stresses will affect the lifetime both by increasing the number of discharges and changing the mechanism of ageing. In many solid dielectrics there is an empirical power law which shows the relation between lifetime and applied stress [37]. Discharge magnitude increases with both height and surface area of a cavity. It has been shown that lifetime reduces with increase of the cavity depth; however lifetime does not change clearly by increase of the cavity surface area [5]. Therefore it is not clear how much discharge magnitude affects the lifetime of the dielectrics [5]. In the case that discharge magnitude is very large; discharge magnitude will shorten the lifetime of insulation.
2.3 Dielectric Spectroscopy 2.3.1
Theory
When a dielectric is subjected to an electric field, the charges bonded to the atoms or molecules of the dielectric rearrange themselves in the electric field. This process is known as polarization. A few mechanisms are known for the process of polarization. At very high frequencies the electrons of an atom slightly displace with respect to the nucleus of the atom, creating a dipole. This is known as electronic polarization. When a material consists of ions exposed to an electric field, the ions displace in the electric field and create a local dipole. This is known
12
as ionic polarization. If a material with permanent dipole subject to an external electric field, the permanent dipoles re-orient in the electric field and create a net dipole. This is dipolar polarization. In the case that a material consist of different phases with different permittivities and conductivities, an external electric field can cause charge accumulation at the interface between the phases. This process is known as interfacial polarization or Maxwell-Wagner polarization. Another possible mechanism is polarization due to hopping charges. This is the case when localized charge can jump from one site to a nearby site inside the medium.
In vacuum the displacement field D is related to electric field E through equation 8.854 10 F/m is the permittivity of vacuum. 2.2, where ε (2.2) If the vacuum is replaced with a dielectric the displacement field increases due to of the dielectric and therefore equation 2.2 has to be modified to polarization equation 2.3. (2.3) The parts of polarization that are very fast such as electronic or ionic polarization together with the vacuum permittivity can be considered as the prompt response and can be treated as a high–frequency permittivity ε , while the rest of polarization which is slow can be kept separated. Therefore equation 2.3 can be also written as equation 2.4. (2.4) For a linear, homogeneous and isotropic material it is possible to obtain a relation between macroscopic polarization and the electric field as it is written in equation 2.5. In equation 2.4, is the dielectric response function of that material. (2.5)
The dielectric response function can be calculated from measurements of polarization or depolarization current (figure 2.1). Equation 2.6 shows the relation that starts at t=0, between current density and a constant applied electric field following a long period with zero electric field. The first term in the right side of the
13
equation 2.6 is the conduction current. The second term is the instantaneous response of the dielectric, where is the delta-function (unit impulse). (2.6)
.
Figure 2.1: Principle of polarization and depolarization current measurement
Considering a sinusoidal excitation, it is possible to rewrite equation 2.5 in the frequency domain as equation 2.7. =
(2.7)
Combining equation 2.7 and equation 2.4 it is possible to derive a complex permittivity term: (2.8) = Dielectric response of material with independent dipoles can be expressed with the Debye response. The Debye relaxation function is given in equation 2.9 and the corresponding complex permittivity can be achieved by taking a Fourier transform of the Debye relaxation function which is shown in equation 2.10. However, in most solids the loss peak is broader than the Debye loss peak. A modified version of Debye response in frequency domain is the Havriliak-Negami function which is
14
shown in equation 2.11. This function does not have any known analytical function in the time domain. (2.9)
/
̃
(2.10)
1
̃
0
,
1
(2.11)
/
In dielectrics where dipoles have significant interaction and hopping charge mechanism may be active, commonly a power-law dependence can be observed. Equation 2.12 shows the Curie-von Schweidler response. , 0
1
(2.12)
A complex permittivity can be calculated by taking the Fourier transform of Curievon Schweidler response, which is shown in equation 2.13. ̃
Γ 1
(2.13)
Γ 1
cos
1
i sin
1
From equation 2.13 it becomes clear that the real and imaginary part of the permittivity becomes two parallel lines in log-log plot for Curie-von Schweidler response. In frequency-domain measurement, conductivity appears in the imaginary contribution, which gives a -1 part of the complex permittivity ( ), with a at low frequencies. More detail discussion about dielectric relaxation slope to can be found in [37, 38].
2.3.2
Diagnostic tool
Dielectric Frequency Response (DFR) and Polarization-Depolarization Current (PC/DPC) measurement are relatively modern diagnostic tools which work based on the dielectric response of the insulation. Traditionally, online measurement of tan is performed at 50/60 Hz on power transformers. Measurement of tan at other frequencies commonly has to be done in offline conditions. Doing the measurement offline has two drawbacks. First an outage of the power equipment is
15
needed and second, the stress during offline measurement is not the same as the stress during normal operation. However there is an advantage of measuring tan for a wide range of frequency. The insulation of a transformer which consists of oil and paper insulation may age with different rate over the transformer lifespan. Measurement at only one frequency may not reveal enough information about the insulation system. Measurement of the tan for a wide range of frequency can be useful for better evaluation of the insulation system. A modern instrument can perform the frequency sweep over a wide range of frequency. Typical values of power factor for transformers and bushings at 50 Hz are categorized by IEEE 621995 which suggests the ranges shown in table 2.1. Table 2.1: Typical values of power factor for transformers and bushings at 50 Hz [39, 40]
Oil insulated power transformers Bushings
Typical value of power factor at 20°C New Old Warning limit 0.2-0.4% 0.3-0.5% >0.5% 0.2-0.3% 0.3-0.5% >0.5%
As mentioned before, moisture (ingress from environment or as a byproduct of ageing process) accelerates the ageing of cellulose. From the diagnostics perspective, evaluation of the moisture content in oil-paper system is therefore an important issue. Moisture distribution between oil and paper eventually reaches an equilibrium at steady state temperature. In this case Karl Fischer titration can be used to evaluate the moisture content of oil, from which the moisture content of paper can be derived by the equilibrium curves [41]. However the equilibrium curves are not very accurate at lower temperatures, and a transformer in operation rarely reaches steady temperature for long time. It has been shown [40, 42–45] that DFR together with the mathematical modeling of the oil-paper insulation provides a good tool for the evaluation of moisture content in the cellulose insulation inside transformers. The XY model is commonly used to model the insulation within transformer. Reference [46] developed a numerical model using finite element method (FEM) to evaluate the changes of the dissipation factor and capacitances. Previous studies show that the effect of moisture on oil-paper insulation is dominant on low frequency part of DFR measurements [40, 47]. Increasing moisture content of the oil-paper insulation results in rapid increase of the real part of permittivity which is mostly visible at frequencies lower than 0.1 Hz and it increases the dielectric loss significantly at frequencies lower than 100 Hz.
16
DFR is affected not only by moisture content of oil-paper insulation, but also by many other factors such as ageing, temperature, structure of the dielectric, acids and chemical composition of the dielectric. A number of investigations have been reported on the effects of these parameters on DFR. A brief summary is given in the following paragraphs. It has been shown [48–50] that the reduction of DP can alter the dielectric characteristics. Investigations performed on thermally aged oil-impregnated pressboard revealed that the real part of permittivity increases with the reduction in cellulose polymer chain length. The real and the imaginary part of the permittivity shift upward in the lower frequency range (1 mHz - 1 Hz) with the ageing condition. Temperature has a large influence on DFR measurement results. Increasing the temperature shifts the dielectric response toward the higher frequencies. A common expression that is used to show the temperature dependency of dielectric response is Arrhenius activation behavior. In the Arrhenius activation equation, a higher activation energy means a stronger temperature dependency of the dielectric response. The activation energy for oil is 0.4 eV and for cellulous is 0.9 eV. This means that a change in temperature, shifts different part of frequency response relative to each other. The activation energy for oil-impregnated paper is measured to be around 0.6 eV [51-55]. If the amount of oil relative to cellulose increases, the interfacial polarization becomes stronger. It has been shown that for a higher percentage of oil to cellulose the loss peak due to interfacial polarization becomes higher [55]. Oil with higher conductivity affects the DFR measurement results of oil-impregnated paper mainly by shifting the dissipation factor toward the higher frequencies [55]. It was shown [56] that formic acid can easily dissolve in paper, while stearic acid easily dissolves in oil. Formic acid has more effect on real and imaginary part of permittivity compared to stearic acid. It is recommended [56] that the influence of acid be considered when calculating the moisture content of oil-paper insulation using DFR. The literature is less rich with the influence of PD on DFR of OIP. In this thesis the effect of PD on DFR of OIP is investigated in detail and compared with other influences.
17
2.4 Breakdown mechanism in solid insulations Breakdown mechanism in solid insulation is a complex phenomenon that locally turns the insulating material into a conductor. When the electric stress increases near to the breakdown stress, the current passing through the insulation increases exponentially. This current is believed to be initiated from electrode emission [37], electron multiplication in the bulk or due to impurities within the solid material which permits electrons to jump from one trap to another (hopping effect) [37]. After an electron is injected into a solid, the electron multiplication is thought to be analogous to the case of gases. Therefore under certain laboratory conditions it is possible that breakdown in solid occurs similarly to gas breakdown. However in real application, factors such as partial discharges, shape of applied voltage, temperature, humidity etc. will affect the breakdown strength. There are a few known mechanisms that may lead to breakdown of a solid insulation [37] which are briefly summarized here.
1- Intrinsic breakdown The term intrinsic breakdown is used when the applied electric field is so high that it can provide sufficient energy to the electrons to cross the forbidden band gap and enter to the conduction band.
2- Electromechanical breakdown Metallic electrodes with insulation between them can be considered as a capacitor. The applied electric field induces charges on the surface of insulation which is proportional to the electric field. These positive and negative charges on each side of insulation attract each other and exert a force on the insulation. Under certain conditions, high electric field may lead to enough force to collapse the solid insulation between the electrodes, which leads to breakdown of the insulation.
3- Electrochemical breakdown Chemical change in insulation may lead to breakdown. For example this change can initiate from oxidation or hydrolysis.
4- Edge breakdown Normally the solid insulation is surrounded by a liquid or gaseous medium. The gases or liquid medium typically has lower permittivity and lower breakdown strength. When determining the breakdown strength of an insulation it is highly possible that breakdown occurs at the edge of the electrode where the weaker
18
medium exists (due to field enhancement and lower breakdown strength of that region). This edge breakdown makes the electrical field highly nonlinear in that region and if it is strong enough it leads to breakdown of the solid insulation. This breakdown voltage is typically lower than the real breakdown voltage of the solid.
5- Thermal breakdown Heat produced by conduction current and polarization within the solid insulation increases the temperature of the insulation. For solid insulation the conductivity and polarization loss typically increase with increased temperature. Therefore this makes a positive feedback, and under certain condition if the loss is high enough it can lead to thermal runaway and cause decomposition of the insulation, and therefore lead to breakdown.
6- Erosion breakdown This happens due to partial discharge activity in contact with solid insulation. A cavity defect is typically filled with gas or liquid that has lower permittivity compared to the solid insulation that surrounds it. This leads a field enhancement inside the cavity and since the breakdown strength of the medium inside a cavity is commonly lower than the solid around it, it leads to partial discharge in the cavity. Over time these PDs degrade the insulation and finally lead to formation of electrical treeing and puncture breakdown.
3 Measurement systems and sample preparation 3.1 PD measurement system In this thesis PDs were detected using the electrical pulse current method. In this method a detection impedance can be placed in series with the test object or in series with a coupling capacitor. When PD occurs an apparent charge is transferred from the coupling capacitor to the test object. This current passes through the detection impedance on which a voltage can be detected. This is shown in figure 3.1.
Figure 3.1: Location of detection impedance, left: in series with test object, right: in series with coupling capacitor
For detecting PD on real transformers, normally a bushing tap is used to record the PD pulses. Figure 3.2 shows the connection used for PD detection on a real transformer.
20
H.V.
PD Calibrator
PD Detector System
Oil
Transformer Winding
Figure 3.2: PD calibration and measurement connection for transformers with capacitive tap on the bushing
Phase Resolved Partial Discharge Analysis (PRPDA) which is a common PD measurement method is used for all PD measurements in this thesis. In this method, the number of PD pulses at each combination of the phase of PD occurrence, and apparent charge are recorded. This was performed by using an ICM-system (Insulation Condition Monitoring), a commercial instrument from Power Diagnostix Systems GmbH [57] that is designed for measurements according to IEC-60270 [58]. Figure 3.3 shows the ICM-system together with the synchronizer and PD calibrator instrument. Figure 3.4 shows a schematic of basic parts and connection to ICM-system. The voltage across the detection impedance is first amplified through a pre-amplifier (better to be placed close to the coupling capacitor), then the amplified signal is sent to the main amplifier which is mounted on the ICM-system. After further amplification through the main amplifier, the signal is sent to the analogue to digital converter (ADC), which registers only pulses above certain magnitude (which can be set by the user). Each phase is registered as an increased count at a particular combination of discrete “channels” of phase and apparent charge. The phase angel of applied voltage at which the pulse occurred is discretized into 256 channels from 0-360 degrees, and the apparent charge is discretized into 256 channels. In this work the charge channels were linearly spaced, with 128 positive and 128 level negative levels. The output of each measurement is therefore a 256*256 matrix of these counts, which is sent through a GPIB port from the ICM-system to a computer. The rows of the output matrix represent the PD amplitude, and the columns represent the phase of occurrence; each element shows the number of PD with specific phase and magnitude. The instrument can record data continuously in “consecutive” mode. The user can set the duration of each measurement, and the pause time between measurements; each measurement is stored as a file with sequential naming. However the ICM has a
21
very limited dynamic range, so that in many cases one has to compromise between losing small PD pulses and saturating the amplifier with the large PD pulses. This is a bit of problem particularly when PD magnitude is changing with time, which case a low amplifier gain would be necessary to ensure avoiding saturation, but as a consequence more pulses would fall below the discrimination level. This issue has to be considered and compensated when variation of PD number is being plotted over time. For the same measurement system, if the capacitances of the test objects (or in general the coupling capacitor, detection impedance, circuit connections) are different, the measured apparent charge can be different. Therefore it is important to calibrate the measurement system before each new set of experiments. A PD calibrator (CAL1D from the Power Diagnostix Systems) that can inject 10 to 1000 pC with positive or negative polarity was used for the purpose of PD calibration. The output of PD calibrator has to be connected across the test object (when the voltage is off) and using the software of the ICM-system it is possible to calibrate the system. More detail about ICM-system can be found in [59, 60].
Figure 3.3: ICM system with synchronizer (left hand side) and PD calibrator (right hand side)
Figure 3.4: Schematic of the basic parts and connections of the ICM system
In most of the experiments in this work, a 100 kV Transformer was used as the voltage source. The transformer input was controlled by a variable transformer in
22
the control unit. For some experiments where higher accuracy for voltage control was needed, a 30 kV voltage amplifier was used as the voltage source. The control input to the amplifier was a function generator (Hewlett Packard 3245A) which generated a 50 Hz sine wave of controllable amplitude. Due to the high level of switching noise that was produced by the high voltage amplifier, a high voltage low pass filter was connected to the amplifier’s output. A schematic of the measurement system when the transformer was used as the voltage source is shown in figure 3.5, and a schematic when the amplifier was used is shown in figure 3.6a. In the setup shown in figure 3.6 the capacitor inside the low pass filter also worked as a coupling capacitor. The low pass filter and its components are shown in figure 3.6b. Using a PD calibrator, it was shown that sensitivity of the measurement system was around to 5 pC. This is normally considered as a low sensitivity however it was enough for these experiments where the PD magnitude was usually much higher than 5 pC. In order to track any change in PD waveform due to ageing, a high bandwidth detection system was also used to capture the PD waveform with high resolution. A 50 ohm resistor was placed in series with the test object. A 1 GHz oscilloscope (Tektronix DPO 4102B) with 5 GSample/s was used to detect the PD waveform. Figure 3.5 shows the connection for this measurement system.
Figure 3.5: A schematic of the PD measurement system with a high voltage transformer as the power supply
23
a
b
L1=L2=80 mH, C1=2C2=500 pF, R1=R2 =40 kΩ Figure 3.6: a) A schematic of the PD measurement system with a high voltage amplifier as the power supply, b) the detail of the low pass filter (capacitors C1 and C2 are ceramic high voltage capacitors).
3.2 Dielectric frequency response measurement In order to investigate the effect of different parameters on dielectric response of OIP, DFR measurements were performed. The complex permittivity was measured by an IDAX 300 dielectric spectroscopy analyzer [61]. This instrument has a builtamplitude. The in voltage source that can apply sinusoidal voltages up to 200 V instrument can measure DFR over a range of frequencies from 10 kHz down to 0.1 mHz. If the measurement has to be performed at voltages above 200 V, an external voltage amplifier can be used as the voltage source, supplied from a 10 V source inside the IDAX 300. The external voltage source used in some experiments was a TREK 30/30 high voltage amplifier which has maximum output of 30 kV. More detail about IDAX 300 can be found in [59]. The impedance of the sample, Z, can be calculated based on the voltage (U) across the sample and the current (I) passing through the sample. The sample can be represented as capacitance with loss. Therefore it is possible to calculate the complex permittivity. Equations 3.1 and 3.2 show the calculation of complex capacitance. Another important parameter is
24
dissipation factor which is defined according to equation 3.3. This parameter has the advantage that the main effects of sample geometry are cancelled. 1
(3.1)
(3.2) tan
(3.3)
Figure 3.7: The IDAX 300 dielectric spectroscopy analyzer
A drawing of the test cell that was used for DFR measurement of OIP is shown in figure 3.8. The electrodes of this test cell are made of stainless steel. The measure electrode (Lo) is separated by a 1 mm air gap from the guard electrode of 2 mm thickness. The measure electrode is loaded by a spring which makes the pressure equal for all measurements. All electrical connections from IDAX 300 to the test cell were made according to figure 3.9. For DFR measurement on oil a standard test cell as described in IEC 60247 [62] was used. A photo of this test cell is shown in figure 3.10.
Figure 3.8: Test cell for measuring dielectric spectroscopy
Figure 3.9: Measurement circuit for measuring dielectric spectroscopy
25
Figure 3.10: Test cell used for measuring dielectric spectroscopy of oil [62]
3.3 Polarization measurement
and
depolarization
current
Measurement of polarization and depolarization current was performed by a Keithley 6514 electrometer and a Keithley 247 DC voltage source which can generate up to 3 kV. For these measurements a test cell, Keithley 6105, (similar to figure 3.7) with a metallic box around it was used. The diameter of the measure electrode is 50 mm separated with a gap of 3 mm from the guard electrode. The metallic box helped to prevent disturbance from moving object close to the measurement system. A schematic of the measurement system is shown in figure 3.11 and a photo of the instruments is shown in figure 3.12. Shielded test cell (Keithley 6105) Coaxial cable
Ajustable DC voltage source (Keithley 247)
Coaxial cable Electrometer (Keithley 6514)
Computer GPIB
Figure 3.11: Schematic of polarization current measurement system
26
Figure 3.12: Polarization-depolarization current measurement setup.
3.4 Sample preparation The oil used in most of the experiments was NYTRO 10XN which is a common oil in transformers. The paper was Munksjö Thermo 70 which has a thickness of 0.1 mm. All the paper samples used in the experiments were impregnated inside a Heraeus VT 5042 EK-GM vacuum oven equipped with a Trivac pump from Leybold. Layers of paper were impregnated with oil by the following procedure [63]: 1- Layers of paper were vacuum (~5 mbar) dried at 120 °C and for 24 hours 2- The temperature of the oven was lowered to 60 °C. The oven chamber was first filled with dry air and then oil that was already pre-heated up to 60 °C was put into the vacuum oven. Both oil and layers of paper were vacuum dried for 24 hour at 60 °C. 3- The vacuum chamber was filled with dry air. The layers of paper were immersed in the oil, and were left to impregnate for 1 day under vacuum at 60 °C. 4- The heater was turned off and layers of OIP were left inside the oil for 24 hours to cool down. In order to minimize moisture ingress from the environment, the layers of paper were stored under oil in a sealed desiccator. Samples of OIP were taken from the desiccator whenever they were needed. No used sample was ever put back into desiccator.
4 Evolution of PDs with time for a cavity embedded between layers of OIP 4.1 Electrical aging of OIP through exposure to PD Continuous PD activity a cavity embedded inside dielectric or on the surface of a solid dielectric can deteriorate the insulation and may lead to a puncture breakdown or sustained surface flashover. During the PD occurrence, high energy particles may break the chemical bonds of the insulation and initiate dielectric deterioration, while discharge byproducts such as ozone or nitric acid can enhance the process [5, 37]. The consequence of deterioration is a reduction of the breakdown strength of solid insulation. In this chapter, the PD behavior in a cavity embedded between OIP layers is investigated. The evolution of PD parameters (number, maximum magnitude and average magnitude of PD) versus time was measured and a comparison of these parameters is made between three cases, i.e. unaged OIP, unaged OIP but with high moisture content and thermally aged OIP samples. Discussion about the mechanism of OIP ageing caused by PD is given at the end of this chapter. The experiments presented in this chapter are all related to an air filled cavity. For the case of an oil filled cavity, which is very common in a transformer, the PD inception voltage is close to the instantaneous breakdown voltage of the OIP. For example, for a sample consisting of 8 layers of paper with a cavity punched in the four middle layers, the inception PD voltage is close to 40 kV if the cavity is filled with oil, but it is below 5 kV if filled with air. Therefore, in the case of an oil filled cavity the small amounts of gas that is formed inside the cavity soon after the PDs become initiated leads to breakdown within very short time, due to the high applied voltage that was necessary to initiate PD in an oil filled cavity. That is why the focus of this chapter is only on a gas filled cavities. A gas filled cavity may form between layers of paper due to bad manufacturing (for example bad bushing impregnation) or in a hot spot region. It also can form due to electrical discharges inside the oil, or ageing of oil and cellulose insulation [64]. Cavities with different diameters and thicknesses were investigated. Even though the cavity geometry is well defined, the amount of gas inside similar cavities may
28
be different. Due to the presence of oil on the surface of paper it is likely that after placing the OIP layers together some amount of oil enters the cavity. In certain cases the oil layer may completely fill the cavity, which can be immediately recognized after applying voltage as PD inception is then much higher than in the expected case. In other cases the amount of air in the cavity varies from one experiment to another. Each experiment presented in this chapter was repeated at least three times with fresh OIP. From the many experiments performed on different OIP samples i.e. unaged, thermally aged and higher moisture content OIP samples, in combination with different cavity geometries, it was found that for this particular defect in OIP the behavior of PD number and maximum magnitude follows a specific trend which can be divided into three main stages. In stage one, immediately after applying the voltage, discharges with large magnitude occur. This stage lasts for a very short time; depending on the cavity geometry it can be a few seconds to a few minutes. In the second stage, both the number and maximum magnitude of PD increase with time. These parameters grow until they reach to a maximum peak at which the next stage begins. In the third stage both the number and maximum magnitude of PD decrease with time, and finally puncture breakdown occurs in the sample. A typical example of PD patterns during the ageing process is shown in figure 4.1.
After 1 minute
After 15 minutes
After 45 minutes
After 75 minutes Continued on next page
29
After 180 minutes
After 306 minutes, one minute before breakdown
Figure 4.1: Change of PD pattern with time (In the two last figures the amplifier gain was changed therefore the scale is different from the others).
For the case where there is only one layer of paper covering the cavity surface (paper I), it is not possible to see any change in the PD patterns even a few minutes before the final breakdown occurs. However, for the case were there are more layers of paper on each side of the cavity, in some cases another discharge pattern appears in the last stage just a few minutes before the final breakdown occurs. A typical pattern related to this particular behavior is shown in figure 4.2.
After 333 minutes
After 336 minutes, a few minutes before breakdown
Figure 4.2: Change of PD pattern during the last few minutes before breakdown
Since the cavity is located in the middle of the sample, the PD pattern could be expected to be symmetric. For most of samples this is what was observed. However in a few cases the PD pattern becomes unsymmetrical in each half cycle during the ageing process. This issue may be due to different oil thickness on each side of the cavity. This unsymmetrical behavior disappears during the ageing time, and the PD pattern becomes symmetric between both half cycles again. This may be due to drying of the oil layers during the ageing. After a few hours of ageing the OIP layers that cover the cavity become dry and the oil layer on the cavity surface vaporizes, i.e. if the sample is opened after a few hours of ageing, the area inside the
30
cavity looks dry. A magnified picture of a cavity surface immediately after the failure is shown in figure 4.3.
Figure 4.3: Change of cavity surface due to PD activity
The ageing process produces gases which are trapped between layers of OIP and can be visually observed if the test cell is opened. The produced gases penetrate between all layers in an area larger than the cavity surface. The produced gases change the real part of permittivity which is discussed in chapter 5. The gas inside the cavity moves to the area outside the cavity, mainly from the edge of the cavity and therefore the area with gases increases. If the sample is opened at the later stage of ageing one can observe that the drying process occurred not only on the first layer that covers the cavity, but also in the same region of all the other layers. This section is usually larger than the cavity area which is shown in figure 4.4. The process of drying continues during the ageing process.
Figure 4.4: Drying of all paper layers in series with the cavity after ageing with PDs
A few possible factors that affect the change of PD parameters with ageing time are described here. The first stage, where larger PD magnitude is observed for a short time, can be attributed to the lack of initial electrons. It is possible that the
31
overvoltage on the cavity due to time lag of initial electron formation leads to a higher magnitude of discharges. The increasing of the PD activity may be due to formation of gases and reactive species created by PD activity together with the change of gas pressure inside the cavity. The gas formed due to oil evaporation can penetrate to the region beyond the cavity and therefore make a wider gas filled area. A wider area is usually accompanied with larger discharges and higher number of discharges. This may be the reason for the increasing part of the curves. Some other researchers claim that the amount of oxygen reduces to almost zero due to discharges inside a cavity. As oxygen is an electronegative gas, this depletion of oxygen may help to increase the PD activity [20, 65]. Another explanation for the change of PD patterns with time may be the change of the surface conductivity [66, 67] of the cavity due to PD activity. The oil layer on the cavity surface has low conductivity at the beginning, but with PD activity the oil conductivity increases, which may lead to higher PD activity. Because over time this oil layer becomes completely dried, in the next stage of ageing, the PD activity decreases again. In the last stage where the PD activity decreases, it is tempting to attribute this change to an increase of the cavity surface conductivity which can reduce the effective voltage across the cavity and therefore reduce the PD activity. However, a simple simulation shows that such a change in conductivity has to be 5-6 orders of magnitude compared to new OIP, in order to have a few percent voltage reductions in the cavity at 50 Hz sinusoidal voltage. Another explanation that can be found from similar research on other materials is the change of PD mechanism from streamer-like discharge to Townsend-like discharges. According to this explanation the reason that PD parameters decrease is because it is not possible to record them with a conventional PD detector. The PD mechanism changes and the PD currents become less sharp, so it is not possible to detect them as pulses [6].
4.2 Limitations of the measurement As was mentioned in chapter 3, the ICM has a limited dynamic range. This means there is a compromise between recording the larger PDs and losing the small PDs. During ageing, PD parameters change with time. In order to record them properly for some cases it was necessary to change the amplifier gain during the ageing process. The change of amplifier gain changes the number of recorded discharges. Figure 4.5 shows variation of PD number for a typical cavity discharge in OIP. The jump in the PD number in region S2 of figure 4.5 is because of changing the amplifier gain. In order to make a comparable curve over time, either region S1 should be mapped to region S2 or vice versa.
32
100
70
S1: 0.545 nC
T3= Thermal fault greater than 700 °C D1= Low energy discharges (sparking)
>
D2= High energy discharges (arcing) DT= Mix of thermal and electrical faults
Figure B.1: Duval’s triangle. The produced gases falls into the region D1 which is corresponded to medium energy discharges Table B.3: Normal and action limit of the gases dissolved in oil (derived from ANSI/IEEE570.104) [65]
Gas description
Key gas concentration (in ppm)
Hydrogen Methane Acetylene Ethylene Ethane Carbon monoxide Carbon Dioxide Total combustibles
Normal limits ()** 1000 80 70 150 35 1000 15000 4630
Potential fault type
Corona, Arcing Sparking Arcing Severe overheating Local overheating Severe overheating Severe overheating
* if the gas values exceed this limit, the frequency of sampling should increase and plan for early outage must be done. ** if the gas values exceed this limit removal of transformer from the service should be considered.
