Identification of Overheating in Transformer Solid Insulation by Polarization Depolarization Current Analysis

Transcription

1 2013 Electrical Insulation Conference, Ottawa, Ontario, Canada, 2 to 5 June 2013 Identification of Overheating in Transformer Solid Insulation by Polarization Depolarization Current Analysis S.A. Bhumiwat Independent High Voltage Diagnostics Consultant P.O. Box Glen Innes Auckland 1743, New Zealand supatra@kea-consultant.com Abstract This article shows how the on-site off-line electrical diagnostic technique, Polarization Depolarization Current (PDC) analysis, identifies and locates overheating / thermal fault in transformer solid insulation. The method is particularly useful for transformers using coated paper of any kinds since the determination of Degree of Polymerization (DP) in paper and the Furanic compounds in oil become impossible. In order to tell more precisely how severe the solid insulation is thermally aged, the Depolarization Index (dp.i.) is here introduced and discussed. Keywords-polarization; depolarization; absorption; conduction; insulation; pressboard; overheating I. INTRODUCTION A. Overheating in transformer solid insulation Heat is the main enemy of insulating paper. Heat breaks the glycosidic bonds of the cellulose thus weakens its mechanical strength and leads to end-of-life. Overheating is the term which refers to abnormal aging of an insulation system due to heat. Overheating in transformer solid insulation can be caused by overloading, supplying fault current, insufficient cooling system, low level of oil, blockage of cooling media, blockage of oil cooling ducts inside the windings, dielectric losses due to high moisture content and/or high level of oxidation products, etc. Insulating paper can sometimes be accidentally overheated during refurbishment. With these varieties of causes, overheating can occur in different areas of solid insulation but only three locations are discussed in this article, which are the insulation system between windings, the high-voltage (HV) ground insulation and the low-voltage (LV) ground insulation. B. Diagnoses of thermal aging & overheating in transformer solid insulation Heat shortens the Degree of Polymerization (DP) of paper, which is the direct measurement of paper life. The DP of a brand new transformer is 1,000 or higher. When the DP reaches , the useful life of insulating paper ends as paper loses its mechanical strength. Unfortunately this measurement of DP requires a paper sample which is difficult to access. DP values of paper taken from different locations are also different. The hot-spot area is usually impossible to reach. As the products from overheating paper are CO, CO 2 and water, the high concentration of carbon oxides, e.g. CO > 500 ppm and CO2 > 5,000 ppm [1], in the Dissolved Gas Analysis (DGA) results of oil from an oil-impregnated paper transformer normally provides the first indication of thermal faults with possible involvement of insulating paper. If CO2/CO is less than 3, the analysis of furanic compounds or a measurement of the DP of paper is advised [2]. Abnormal increase of 2- furfuraldehyde (2-FAL) in oil confirms thermal faults or overheating in insulating paper. However, the generation of carbon oxides and furanic compounds can be much less when thermally upgraded paper is applied together with nonthermally upgraded paper in the same transformer. Thus makes the interpretation of test results difficult and inaccurate. Recent researchers [3] and [4] are working on Methanol in oil as another chemical aging marker for transformer solid insulation. Since the previous decade there has been an increasing use of epoxy-diamond-pattern-coated paper in distribution and subtransmission transformers, with the purpose of increasing mechanical strength. Epoxy resin is applied to thermallyupgraded paper in the pattern of diamond dot. Although oil can still penetrate among the diamond dots, the DP measurement of paper is not possible. Other situations when DP of paper cannot be applied include the application of varnish in some old transformers and the application of enamel to new rewound coils after transformer failure caused by corrosive sulfur. This paper introduces an electrical, off-line dielectric response technique [5]-[8], Polarization Depolarization Current (PDC) Analysis, which can identify and locate overheating in any type of transformer solid insulation. A transformer which has oil cooling ducts inside the windings needs to be tested without oil (e.g. when the oil is drained after faults or during transformer refurbishment) but a transformer which has no oil ducts can be tested with oil filled. In this article all PDC results are from a Swiss-made PDC-Analyzer- 1MOD (by ALFF Engineering) which provides measurement results in time-domain together with the evaluation results in frequency domain. But only the time-domain measurement results are used in the identification of aging types in transformer solid insulation, as already discussed in [9]-[10]. Other applications of PDC analysis are in [11]-[13] /13/$ IEEE 449

2 II. HOW PDC IDENTIFIES AND LOCATES OVERHEATING IN TRANSFORMER SOLID INSULATION A. Identification of aging in solid insulation by PDC shape The PDC-Analyzer provides measurement results of polarization (or charging) current in addition to depolarization (or discharging) current in order to identify conduction and polarization in electrical insulation. During charging with a constant direct voltage, the current consists of the steady-state conduction current (caused by conductive contaminants e.g. surface humidity, free water, wet particles, metal particles, carbon dust, etc.) and time-dependent absorption current (caused by deterioration products or products from chemical reaction e.g. oxidation products, by-products from overheating or partial discharges, etc.). During discharging, the current consists of absorption current only since conduction current does not exist when external power source is removed. So it is the depolarization current that identifies aging which causes chemical change in the insulating materials. When transformer solid insulation has very low conductive contaminants, the polarization current (I pol) and the depolarization current (I depol) will be nearly equal for about one-tenth of the charging time such as the PDC shape of solid insulation between windings of a 20 MVA, 33/11 kv, transformer KEA-01, age 36, after refurbishment in fig. 1. The evaluation of moisture in pressboard from the PDC-Analyzer is % (was 4% before refurbishment). Moisture in the bulk solid insulation (without surface humidity) increases absorption current only. Regardless of current amplitude, the PDC shape looks quite straight in log-log scale (no bending or crook) which means neither overheating nor severe deterioration of solid insulation. This transformer has had normal operation with light load throughout its service life. B. A case of overheating in the pressboard between windings Fig. 2 presents PDC measurement results of a 2 MVA, 33/11 kv sealed transformer KEA-02, age 3, after removed from service due to very high concentration of fault gases in Dissolved Gas Analysis (DGA) results and after the oil was drained (for internal inspection). The very high concentration of H 2 as shown in table I was from chemical reaction of the radiator galvanized steel which also caused solids by-products or debris inside the radiator and led to poor cooling system or overheating. The DGA results revealed overheating in both oil and paper. But no furanic compounds were detected. Some other oil test results are included in table 1 for reference. In this transformer epoxy-diamond-pattern-coated paper is used in 1.E-12 PDC at 100V, 15 o C Figure 1. PDC measurement results of the solid nsulation between windings of Transformer KEA-01, age 36, after refurbishment before oil refilling TABLE I. OIL TEST RESULTS OF TRANSFORMER KEA-02 (WITH COATED PAPER) Description Test 1 Test 2 Days in service 1,003 1,036 Oil temperature ( o C) Dissolved Gas Analysis (unit: ppm) Hydrogen (H 2) 18,073 18,390 Oxygen (O 2) 4,155 2,328 Nitrogen (N 2) 43,596 42,307 Carbon monoxide (CO) Carbon dioxide (CO 2) 1,699 2,018 Methane (CH 4) 1,650 1,659 Acetylene (C 2H 2) < 1 < 1 Ethylene (C 2H 4) 4 5 Ethane (C 2H 6) Furanic compounds (unit: ppb) 2-furfural (2-FAL) < 10-5-hydroxy-methyl-furfural (5HMF) < 10-2-acetylfuran (2-ACF) < 10-5 methyl-2-furaldehyde (5-MEF) < 10-2-furfuryl alcohol (2-FOL) < 10 - Other paper aging markers (unit: ppb) Methanol - - Ethanol - - Other oil test results Moisture content (ppm) 6 - Conductivity (IEC61620) at 20 o C (ps/m) Interfacial tension (dynes/cm) Acid number (mgkoh/g) Color number (relative) Oxidation inhibitor, DBPC (%wt.) < PDC at 100V, 24 o C (HV-LV) PDC at 100V, 24 o C (HV-G) PDC at 100V, 24 o C (LV-G) Figure 2. PDC measurement results of Transformer KEA-02 without oil: ` (Middle) - HV ground insulation (HV-G) and (Bottom) - LV ground insulation (LV-G) 450

3 both high voltage (HV) and low voltage (LV) windings, but normal pressboard is used as the insulation between HV and LV. The PDC measurement results in the top chart of fig. 2 shows I depol of the insulation between windings (HV-LV) has prominently crooked shape which refers to overheating. Since conduction current of HV-LV is very low, the crooked shape of absorption current is obvious in both I pol and The PDC shape of HV ground insulation (HV-G) and LV ground insulation (LV-G) in the middle and bottom charts of fig. 2 looks quite straight in log-log scale (no bending or crook) which means quite normal. Conductive contaminants in HV-G and LV-G can be noticed by the results of I pol that the steadystate conduction current exists although the amplitude is not high. C. A case of overheating in the LV ground insulation Fig. 3 presents PDC measurement results of an old 1 MVA, 33/11 kv free-breathing transformer KEA-03 after overloading. The prominent crook of PDC shape is obvious on I depol of LV-G but neither the insulation HV-LV nor HV-G. This means only LV ground insulation was overheated from the overloading. The transformer KEA-03 was manufactured in 1959 and has no oil cooling ducts inside the windings. So there is no need to remove the oil for PDC analysis in the assessment of solid insulation. The oil test results in table II show very high increase of CO, CO 2 and furanic compounds in test 2 and test 3 after overloading compared with test 1 which was before overloading. The results mean the insulating paper was severely overheated. Test 3 in table II also includes the Methanol & Ethanol results (for reference only), which have high potential to become new paper aging markers. Since water is also one of the by-products of paper heating, it can be seen in the same table that moisture-in-oil increased from 19 ppm in test 2 to 40 ppm in test 3 after continued overloading. The high moisture in the insulation system can also be confirmed by the big difference between I pol and I depol in the PDC measurement results of HV-LV, HV-G and LV-G in fig. 3. D. A case of overheating in the HV ground insulation Fig. 4 present PDC measurement results of a17.5 MVA transformer KEA-04, age 28 after the oil was drained in the workshop in order to replace the leaky gasket at top cover and to inspect the cause of overheating in oil as appeared in DGA results in table III. It is obvious from the PDC results that the crooked shape of I deploy is at HV ground insulation. In addition, conductive contaminants can be easily noticed in I pol of HV ground insulation but are at high level for LV ground insulation. Overheating in the HV ground insulation is unusual since HV winding is normally the outer winding where there is always sufficient amount of oil for cooling. But this transformer suffered ground fault on HV side phase C nearly 12 years ago and was repaired. The oil was also replaced. But the internal inspection after PDC test revealed solid debris remained on phase C. The microscopic evaluation results identified most of the debris were organic fibres and inorganic crystalline dirt. About 10% of debris was metallic particles. TABLE II. OIL TEST RESULTS OF THE OLD TRANSFORMER KEA-03 Description Test 1 Test 2 Test 3 Days in service 17,529 18,514 18,782 Oil temperature ( o C) Dissolved Gas Analysis (unit: ppm) Hydrogen (H 2) Oxygen (O 2) 27,110 11, Nitrogen (N 2) 60,930 58,368 60,597 Carbon monoxide (CO) Carbon dioxide (CO 2) 1,562 12,584 8,802 Methane (CH 4) Acetylene (C 2H 2) < 1 < 1 < 1 Ethylene (C 2H 4) Ethane (C 2H 6) < Furanic compounds (unit: ppb) 2-furfural (2-FAL) 1,320 2,600 3,550 5-hydroxy-methyl-furfural (5HMF) < acetylfuran (2-ACF) < 10 < 10 < 10 5 methyl-2-furaldehyde (5-MEF) < furfuryl alcohol (2-FOL) < 10 < 10 < 10 Other paper aging markers (unit: ppb) Methanol Ethanol Other oil test results Moisture content (ppm) Conductivity (IEC61620) at 20 o C (ps/m) Interfacial tension (dynes/cm) Acid number (mgkoh/g) Color number (relative) Oxidation inhibitor, DBPC (%wt.) - < PDC at 100V, 22 o C (HV-LV) PDC at 100V, 22 o C (HV-G) PDC at 100V, 17 o C (LV-G) Figure 3. PDC measurement results of Transformer KEA-03 filled with oil but the transformer has no oil duct: ` (Middle) - HV ground insulation and (Bottom) - LV ground insulation 451

4 TABLE III. OIL TEST RESULTS OF TRANSFORMER KEA-04 TABLE III. OIL TEST RESULTS OF TRANSFORMER KEA-04 Description Test 1 Test 2 Description Days in service Test 1 8,938 Test 2 9,771 Days Oil temperature in service ( o C) 8, , Oil Dissolved temperature Gas Analysis ( o C) (unit: ppm) Dissolved Hydrogen Gas (H 2) Analysis (unit: ppm) Hydrogen Oxygen (O(H 2) 2) 2, , Oxygen Nitrogen (O(N 2) 2) 83,000 2,000 99,000 5,500 Carbon Nitrogen monoxide (N 2) (CO) 83, , Carbon monoxide dioxide (CO (CO) 2) 3, , Carbon Methane dioxide (CH 4) (CO 2) 3, , Methane Acetylene (CH 2H 4) 2) 160 < < 1 Acetylene Ethylene (C (C 2H 2H 4) 2) < 71 < 331 Ethylene Ethane (C (C 2H 2H 6) 4) Furanic Ethane compounds (C 2H 6) (unit: ppb) Furanic 2-furfural compounds (2-FAL) (unit: ppb) 70-2-furfural 5-hydroxy-methyl-furfural (2-FAL) (5HMF) < hydroxy-methyl-furfural 2-acetylfuran (2-ACF) (5HMF) < 10-2-acetylfuran 5 methyl-2-furaldehyde (2-ACF) (5-MEF) < furfuryl methyl-2-furaldehyde alcohol (2-FOL) (5-MEF) < 10 - Other 2-furfuryl paper alcohol aging markers (2-FOL)(unit: ppb) < 10 - Other Methanol paper aging markers (unit: ppb) - - Methanol Ethanol - - Other Ethanol oil test results - - Other Moisture oil test content results (ppm) 9 9 Moisture Conductivity content (IEC61620) (ppm) at 20 o C (ps/m) Conductivity Interfacial tension (IEC61620) (dynes/cm) at 20 o C (ps/m) Acid Interfacial number tension (mgkoh/g) (dynes/cm) Acid Color number (mgkoh/g) (relative) Color Oxidation number inhibitor, (relative) DBPC (%wt.) Oxidation inhibitor, DBPC (%wt.) PDC at 100V, 28 o C (HV-LV) PDC at 100V, 28 o C (HV-G) PDC at 100V, 28 o C (LV-G) Figure 4. PDC measurement results of Transformer KEA-04 without oil: ` (Middle) - HV ground insulation and (Bottom) - LV ground insulation 452 III. HOW TO ENSURE ACCURATE AND SAFE MEASUREMENT In order to use depolarization current to judge overheating in solid insulation accurately, it is important that the test instrument is able to monitor remaining charges (through depolarization current) in the insulation system before any voltage application. The discharging should be applied until the remaining current (depolarization current) is very low and has constant amplitude upon time before actual measurement starts. Correction should then be made to all measurement data using the constant value of the remaining current to obtain precise results. Charging time or polarization time is another important issue since it influences the depolarization current around final time. It is recommended that charging time is not less than 1,000 seconds. The charging time of 2,000 seconds is found to be a good compromise to ensure decisive results. When testing a transformer without oil for the assessment of aging in solid insulation, special care should be taken that the test voltage amplitude is very low in order to ensure negligible stress to the solid insulation. Authors suggested in [9] that the test voltage should not be higher than 100V. IV. DEPOLARIZATION INDEX In the above case studies, overheating in transformer solid insulation was identified by visual notice of prominently crooked shape of the depolarization current (or the major change of slope to steeper) which may seem to have low precision of diagnosis. But the whole shape of PDC provides full information of the information system measured e.g. how high conductive contaminants are (the earlier the steady component of I pol starts, the higher level of conductive contaminants are), when the depolarization current starts to change the slope to steeper, or whether I depol within 10 seconds has steeper slope than the later time scales (which can mean problem of solid insulation closer to the conductor), etc. When the insulation system has very low level of conductive contaminants, the high values of Polarization Index (P.I.) between 60 and 600 seconds e.g. P.I. > 7 can be used to confirm overheating in solid insulation such as in the case of KEA-02: HV-LV which has P.I But when the insulation system has high level of conductive contaminants, the index from I pol is useless since conduction component has strong influence in lowering P.I. closer to 1. Using the index from I depol to judge aging in solid insulation is recommended since I depol represents absorption current only. The author would like to introduce the term Depolarization Index (dp.i.) for the index from I depol but it is too early to tell if the high dp.i. between 60 and 600 seconds is the most suitable time interval in the judgment of overheating and how high dp.i. should be. Different causes of overheating or thermal faults can lead to different time scale of the prominent crook in PDC shape. Other aging type(s) can be combined to overheating which can lead to more crooks of the PDC (or more major changes in the slope of absorption current). Table IV presents the values of dp.i. at different time intervals for all transformers in the above case studies as a numerical reference.

5 TABLE IV. DEPOLARIZATION INDEX OF TRANSFORMERS IN CASE STUDIES Description KEA-01 KEA-02 KEA-03 KEA-04 HV-LV Charging time (s) 10,000 1,000 2,000 2,000 P.I. (60 & 600 s) dp.i. (1 & 10 s) dp.i. (10 & 100 s) dp.i. (20 & 200 s) dp.i. (60 & 600 s) HV-G Charging time (s) - 1,000 1,400 2,000 P.I. (60 & 600 s) dp.i. (1 & 10 s) dp.i. (10 & 100 s) dp.i. (20 & 200 s) dp.i. (60 & 600 s) LV-G Charging time (s) - 1, ,000 P.I. (60 & 600 s) dp.i. (1 & 10 s) dp.i. (10 & 100 s) dp.i. (20 & 200 s) dp.i. (60 & 600 s) V. CONCLUSION The depolarization current represents the characteristic of transformer solid insulation during PDC measurement of transformer without oil as long as the test voltage is very low. The transformer which has no oil cooling ducts between windings can be tested with oil filled. Overheating in transformer solid insulation is identified by the prominently crooked shape of the depolarization current. In another word, the slope of the depolarization current has a major change to steeper towards lower current amplitude. Remaining charges stored in the solid insulation and charging time influence the depolarization current. Advice is given how to ensure accurate and safe measurement. It is by PDC analysis that overheating in solid insulation can be localized. The case studies in this article demonstrate that overheating in paper of an in-service transformer can occur in the insulation between windings and/or ground insulation of any winding. From the author s PDC database, overheating in HV ground insulation is rare for a core-type transformer which never has faults. As usually an outer winding, the surface area of paper in contact with oil is larger than the other locations. When a transformer is under heavy load or supplying fault current, the worst area of paper overheating can be LV ground insulation or the insulation between windings or both, depending on the design, type of cooling system and whether the oil cooling ducts are blocked by the accumulation of oxidation products. Finally, the Depolarization Index (dp.i.) is firstly introduced in this article to ensure judgment of overheating in transformer solid insulation is accuratetly made. Higher value of dp.i. (e.g. > 6) means more severity of overheating. Although dp.i. between 60 and 600 seconds provides better sensitivity in many cases, the selection of time interval can vary from case to case, depending on location and type of thermal faults. ACKNOWLEDGMENT Many thanks to my clients from whose transformers the data presented here were collected. Also special thanks to Dr. J. Jalbert from Institut de recherche d Hydro-Quebec for providing the test result of Methanol & Ethanol in table II as a reference. REFERENCES [1] IEEE Std. C IEEE Guide for the Interpretation of gases generated in oil-immersed transformers. [2] IEC60599 Mineral oil-impregnated equipment in service Interpretation of dissolved and free gases analysis. [3] Y. Denos, A. Tanguy, P. Guuinic, R. Gilbert, J. Jalbert, P. Gervais, Ageing diagnosis by chemical markers, influence of core-type and shell-type technology, " in CIGRE 2010 Session, Paper no. A [4] A. Schaut, S. Eeckhoudt, Identification of early-stage paper degardation by Methanol in CIGRE 2012 Session, Paper no. A [5] J. Alff, V. Der Houhanessian, W. S. Zaengl and A.J. Kachler, "A Novel, Compact Instrument for the Measurement and Evaluation of Relaxation Currents conceived for On-Site Diagnosis of Electrical Power," in 2000 IEEE International Symposium on Electrical Insulation, Anaheim, USA, April 2-5, 2000, pp [6] W. S. Zaengl, Dielectric spectroscopy in time and frequency domain for HV power equipment, Part I: Theoretical considerations. IEEE EI Magazine, vol. 19 no. 5, September/ October 2003, pp [7] W. S. Zaengl, Applications of dielectric spectroscopy in time and frequency domain for HV power equipment. IEEE Electrical Insulation Magazine, vol. 19 no. 6 November/December 2003, pp [8] CIGRE Task Force , Dielectric response methods for diagnostics of power transformers. IEEE EI Magazine, vol. 19 no. 3, May/June 2003, pp [9] S. Bhumiwat, S. Lowe, P. Nething, J. Perera, P. Wickramasuriya, P. Kuansatit, Performance of oil and paper in transformers based on IEC61620 and dielectric response techniques. IEEE EI Magazine, vol. 26 no. 3, May/June 2010, pp [10] S. A. Bhumiwat, Advanced Applications of Polarisation / Depolarisation Current Analysis on Power Transformers, in 2008 IEEE International Symposium on Electrical Insulation, Vancouver, Canada, June 2008, Paper no [11] S. Bhumiwat, P. Stattmann Quality assurance after transformer refurbishment by means of polarisation depolarisation currents analysis, in 2003 IEEE Bologna Power Tech Conference, Bologna, Italy, June 23-26, 2003, Paper no. 90. [12] S. A. Bhumiwat, P. Phillips, Verification of on-site oil reclamation process by means of polarisation depolarisation currents analysis, in 2004 IEEE International Symposium on Electrical Insulation, Indianapolis, IN, USA, September 19-22, 2004, pp [13] S. A. Bhumiwat, Insulation condition assessment of transformer bushings by means of polarisation / depolarisation current analysis, in 2004 IEEE International Symposium on Electrical Insulation, Indianapolis, IN, USA, September 19-22, 2004, pp