Abstract: Dielectric properties of bushings were inves- tigated by measuring of polarisation- and depolarisation currents PDC. Network simulations were based on spatial discretisation and equivalent circuits represent- ing material system properties. Different scenarios were simulated and exemplarily verified by measurements: Conductive surface layers, water in the body and in the surface of the core (oil impregnated paper OIP), ageing, non-uniform distribution of water, measurements with surface bandages. Normally currents from the measur- ing tap of a bushing are assigned to the insulating core, which is only justified if parasitic currents do not have access to free ends of grading foils. With surface ban- dages either connected to ground or to diagnostic volt- age, measured currents give lower and upper limits. Guard ring bandages are used to find improved estima- tions of the current through the core. It is proposed to calculate conductivities with a new charge difference method CDM. Results: (1) Environmental influences can be excluded by means of bandages. Conductivities are insensitive against parasitic currents. (2) Selective information about insulation parts can be extracted. (3) Conductivity is related both to moisture and ageing. (4) The analysis of initial polarisation currents is the first known dielectric method for identification of aged OIP- bushings at room temperature. 1 INTRODUCTION A significant percentage of transformer failures is related to bushing insulation defects. In the future an increasing number of progressively ageing oil-impreg- nated OIP-bushings will be subjected to increasing thermal stresses due to increasing power flow require- ments and overloading conditions. This can result in accelerated ageing, enhanced dissipation factors and thermal instabilities. Additionally water increases dissi- pation and ageing and reduces dielectric strength. If partial breakdowns between grading foils occur, the overall dielectric strength is seriously affected. In case of strategically important bushings there is a strong need for reliable condition assessment by dielec- tric diagnosis of partial breakdowns, ageing condition and water content. Todays dielectric diagnosis of bushings is based on offline power frequency measurements of capacitance C and dissipation factor tan at the measuring tap which is connected to the outermost grading foil, Fig. 1. The capacitance C is a sensitive quantity for the detection of partial breakdowns between grading foils, but it does not change significantly during slow ageing processes in oil impregnated paper OIP. During offline measure- ments, when dissipation factors tan are measured at ambi ent temperature, the values are low and insignifi- cant, even for strongly aged or wet OIP-insulation. Dis- sipation factors at service temperatures above 50 C remain unknown, although they might be high and dan- gerous. It was found that other dielectric measurements can give much better i ndications, even at ambient tem- perature. In general dielectric measurements can be performed as frequency domain analysis FDA or as time domain analysis with polarisation and depolarisation currents PDC. For linear systems both methods are mathemati- cally equivalent [1], [ 2]. The authors decided to use PDC analysis because of the following reasons: Fig. 1: Dielectric measurement at the measuring tap of a bushing with access of parasitic surface currents to the free edges of the grading foils in the middle of the grading contour. (1) Basically PDC measurements give step respon- ses containing the whole system information. (2) Time domain signals can be observed easily, explained by physical models (ion movement, polarisation processes) and described clearly (ion transit times, equivalent cir- cuits). (3) Currents at different times are related to dif- ferent influences [3] (e.g. oil quality, ageing, moisture). (4) The PDC analysis has successfully been applied to transformer diagnosis [4], [5], [ 6] and bushing diagnosis [5], [7]. (5) PDC allow to calculate d.c. conductivities containing information about moisture and ageing [8]. (6) Procedures have been developed to extract relevant information from very short measurements. 2 MEASUREMENTS AND SIMULATIONS 2.1. Dielectric Measurements on Bushings Normally it is assumed that dielectric measurements on bushings can directly be related to the OIP bushing core. This is not always true: Sometimes leakage cur- rents have access to the grading foils, Fig. 1, and can be responsible for negative dissipation factor measure- ments [ 9]. It is concluded that polarity reversals during polarisation current measurements are caused by leakage currents. For explanation, a bushing core is basically de- scribed by two capacitances C a and C b and one grading foil in-between, Fig. 2 and 3. A conductive path is as- sumed from ground or h.v. side to the grading foil (left and right). In the frequency domain the influence of the conductive path can be described by a phase shift between measured current I b and applied voltage U which can be bigger or smaller than 90, Fig. 2 (left and right). Thereby the dissipation factor appears to be negative or positive. Notice: It is an apparent effect, in reality the ideal capacitance does neither produce nor dissipate energy [10] [11], [12]. For the description in the time domain, additional elements are introduced to the equivalent circuit in order to consider polarisation and conductivity, Fig. 3. The influence of the conduc- tive path can be described analogously by a reduction or an enhancement of the measured current i b (t), Fig. 3 (left and right). A conductive path to ground can result in a temporary dischar ging of C b , in a negative current i b
and in two polarity reversals, Fig. 3 (left). It is concluded that dielectric measurements on bushings can be disturbed by parasitic currents to an unknown extend. 2.2. Simulation of Bushings The behaviour of the bushing with and without para- sitic currents was simulated with a network model in order to understand the impact of external influences and in order to look for measuring procedures which allow to perform correct dielectric measurements.
Fig. 2: Frequency domain: A conductive path from a bushing grading foil to ground or h.v. (left and right) results in appar- ent reduction or enhancement of dissipation factor tan . The bushing core is described by two series capacities C a and C b .
Fig. 3: Time domain: A conductive path from a bushing gra- ding foil to ground or h.v. (left and right) results in reduction or enhancement of polarisation current. Even polarity rever- sals are possible (left). The bushing is described by C a , C b and additional elements (polarisation and conductivity). The nodes of the model were meshed in axial and radial orientation [10]. Each connecting element con- sisted of a capacitance (replacing permittivity), a resis- tance (replacing conductivity) and RC-elements (re- placing different polarisation processes), Fig. 4. These equivalent circuits, describing the local materials, were derived from real PDC measurements on material sam- ples for all materials used in a bushing [11], [13], [14], [15], Fig. 5. In the case of nonlinear oil, a nonlinear algorithm was used. The results of simulations are in good agreement with measurements [ 10]. The simula- tion model was used to calculate transient currents dur- ing PDC measurements.
Fig. 4 : 123 kV transformer bushing with OIP core (left), spacial discretisation (middle and bottom) and attribution of equivalent circuits representing insulating materials (right).
Fig. 5: Dielectric system responses for different insulating materials at E = 0.1 kV/mm and room temperature (top) and modelling of linear and nonlinear materials (bottom). 3 INFLUENCE OF PARAMETERS The influences of different parameters on dielectric measurements on bushings were investigated with PDC measurements and with network simulations. PDC measurements were performed with a PDC analyser with a high dynamic range from 1 pA to 1 mA [16]. The simulation quality was checked by comparison with measurements. Diagnostic situations, which could not be investigated by measurements, were simulated. 3.1. Measurements with Bandages External influences as described in Fig. 3 - were investigated by a set of three measurements at the measuring tap of a bushing, Fig. 6: (1) The traditional measurement without bandages, (2) a measurement with a circumferential bandage in the middle of the grading contour at ground potential and (3) a measurement with the same bandage at diagnostic voltage (h.v.). Measurement (1) gives an estimation of the current through the bushings OIP core. Measurements (2) and (3) impose extreme values of parasitic surface currents as assumed in Fig. 3. There- fore the measured currents can be interpreted as lower and upper limits of the current through the bushings OIP core, Fig. 6 (top). The bandages can be called worst (extreme) case bandages showing the sensitivity of the bushing to parasitic currents. The behaviour of the bushing with and without ban- dages was simulated with a network model. The results of simulations are in good agreement with measure- ments, Fig. 6 (top and bottom). The simulation model was used to optimise the po- sition of the bandage. It was found, that a grounded bandage above the edge of the outermost layer on the air side of the bushing collects all relevant leakage currents on the air side of the bushing. It can be called guard ring bandage protecting the measurement. The current taken from the measuring tap is therefore identical with the current through the bushings core.
Fig. 6: PDC measurements without and with bandages in the middle of the grading contour. Bandages at h.v. and ground give upper and lower limits for polarisation currents through the bushings core (worst case bandages), measurements (top) and simulations (bottom).
Fig. 7: Polarisation currents measured on a severely aged 420 kV OIP bushing, measured at RT and 1 kV. Bandages in the middle of the grading contour give lower and upper limits for the polarisation current through the bushings core. If the bushing is new, these limits are clearly separated, Fig. 6 (top). In the case of severely aged bushings, currents are strongly enhanced and leakage currents are negligible, Fig. 7. Upper and lower limits are close together and traditional measure- ments without bandages are considered to be correct. 3.2. Surface Conductivity on the Air Side Measurements with conductive surface layers of in- creasing length on the air side of the bushing were real- ized by conductive bandages, simulations were made with increasing lengths of surface resistances, Fig. 8 (top and bottom). Surface conductivities result in trans- ient polarisation currents with polarity reversals, both in measurement and simulation. This is in accordance with theory, Fig. 3. Furthermore, there is a good agreement between measurements and simulations which shows that simulations can be used for parameter studies. It is concluded that transient currents are sensitive to external influences, e.g. caused by wet and dirty porce- lain surfaces, but long term currents and conductivities are measured correctly. This means that measurements should be handled with care, if surfaces are not clean and dry. In these cases the analysis should be based on current end values. Further improvement can be made by application of bandages as described in chapter 3.1. 3.3. Surface Conductivity on the Transformer Side It was further investigated whether the conductivity of the transformer oil or conductive deposits on the bushings transformer side might cause leakage currents which have access to the grading foils and which can influence the measured currents. Measurements with bandages and simulations with surface resistances were made for different lengths of the surface layers, Fig. 9 (top and bottom). Both the measurements and the simulations give same results: There is no influence for long term values because of the high resistive epoxy housing insulator on the transformer side. For shorter times, the transient polarisation currents are influenced by a capacitive coupling from the bushing surface to grading layers on the transformer side.
Fig. 8: Polarisation currents at the measuring tap of a 123 kV OIP bushing with conductive surface layers on the air side: Measurements with conductive bandages (top) and simula- tions (bottom).
Fig. 9: Polarisation currents at the measuring tap of a 123 kV OIP bushing with conductive surface layers on the trans- former side: Measurements with conductive bandages (top) and simulations (bottom). Therefore it is advisable to base the analysis on long term values, which was already concluded in chapter 3.2. Current end-values and conductivities can be cal- culated with a new charge difference method CDM from the difference of polarisation and depolarisation charge [11], [13], also described in an ISH paper [15]. Another result of these investigations is that power frequency dissipation factor measurements are influ- enced in a similar way, but to a smaller extend [10]. Nevertheless, frequency domain analysis FDA of di- electric measurements will be subject to the same influ- ences as described for PDC measurements. 3.4. Water in Core and Surface Layers Water in an OIP bushing core increases the polari- sation currents, Fig. 10 (top and middle). The relations were investigated on material samples and described in further papers [14], [15]. The influences on dielectric measurements on bushings were investigated by means of simulations: If the bushing core is dry, the influence of parasitic currents is comparatively remarkable, as it can be seen from the measurements with bandages, Fig. 10 (top). If the core is wetted, the influence of parasitic currents is negligible, Fig. 10 (middle).
Fig. 10: Simulation of polarisation currents at the measuring tap of a 123 kV OIP bushing with and without bandages (outer and inner curves). Top: Dry OIP core with a water content w = 0.74 %. Middle: Wetted OIP core with a water content w = 5 %. Bottom: Dry OIP core (w = 0.74 %) with wetted surface paper layers above grading contour (w = 5 %). Furthermore, it was investigated, if inhomogeneous distribution of water in the core could be detected, but unfortunately significant differences could not be de- tected. Therefore dielectric measurements give integral information about bushing core properties only [17]. Nevertheless, if the outer paper layers of the core are wetted, measurements with bandages give a signifi- cantly wider separation of the three curves, see Fig. 10 (bottom) in comparison with Fig. 10 (top). This can be explained by a reduced radial resistance of wetted OIP layers between grading layers and bushing surface with bandages at ground or diagnostic voltage. It is concluded, that diffusion of water from the out- side into the bushing core can be detected from the separation of measurements with bandages at ground and diagnostic voltage. A dry core can be detected from the low polarisation currents during measurements without bandage. 3.5. Ageing Both ageing and water result in increasing polarisa- tion currents, Figs. 7 and 10. It was found, that ageing of OIP bushings is mainly related to the ageing of the oil component [11], [14], [15]. Measurements on service aged bushings show, that the severely aged bushings no. 1 and 5 (see Fig. 11) can clearly be detected from high initial polarisation currents, even in room temperature measurements [10]. All of the bushings had low power frequency dissi pation factors tan at room temperature, but bushings no. 1 and 5 were thermally instable above 70 C. They had already been overheated in service which resulted in gassing and partial discharges. Fur- thermore, ageing and wetting can be distinguished from the gradient of the polarisation current curve. It is concluded that PDC analysis is the first method which is able to detect severely aged bushings from dielectric measurements at ambient temperature. The influences of ageing and water can be distinguished, if initial and end values of polarisation currents are con- sidered.
Fig. 11: PDC measurements on identically designed 420 kV OIP bushings, differently aged in service, measurements at RT and 1 kV. (1), (5) severely aged (2), (3) normally aged (4) new OIP core High polarisation currents indicate severe ageing already at RT 4 CONCLUSIONS The behaviour of bushings during dielectric meas- urements was investigated and compared with the re- sults of network simulations. The following conclusions are drawn: (1) The PDC analysis is a powerful tool for the evaluation of the dielectric properties of bushings, but bushing core properties can not be measured correctly, if parasitic currents have access to the grading foils. Thereby dissipation factors (frequency domain) and polarisation currents (time domain) can be increased or reduced apparently. (2) It is proposed to use conductive bandages above the grading contour to determine the sensitivity of the bushing against external influences. With two additional measurements (bandage at ground and at diagnostic voltage) lower and upper limits can be found for polari- sation currents (worst/ extreme case scenarios). Net- work optimisation showed that the pure bushing core current is measured, if a grounded bandage is situated above the edge of the grounded layer (guard ring effect). (3) Dielectric measurements on severely aged or wetted OIP bushings are not sensitive to parasitic sur- face currents, but generally they have to be considered: a) Conductive deposits on the air side of the bushing surface have a strong influence on transient polarisation currents but not on long term end values and conduc- tivities. It is proposed to clean and dry surfaces, to measure with bandages and to prefer evaluation of cur- rent end values. b) Parasitic currents on the transformer side are blocked by a high resistive housing insulator, but only for long measuring times. Therefore again it is proposed to consider long term current and conductivity values. (4) It is possible to distinguish the properties of the core and the surface layers by PDC measurements with and without bandages from the relative distance be- tween the curves. (5) Severe ageing of OIP especially causes high val- ues of polarisation currents in an early phase during some hundred seconds after beginning of polarisation. Thereby severely aged OIP bushings (with critical di - electric losses at service temperature) could be detected and distinguished from wetted insulation by PDC meas- urements already at room temperature. The PDC analysis is the first method for the identifi- cation and separation of aged and wetted OIP insulation at ambient temperature. 5 ACKNOWLEDGMENT The authors gratefully acknowledge the financial support of this work by the Deutsche Forschungsge- meinschaft DFG (KU 1384/3-1 and -2). 6 REFERENCES
[1] W.S. Zaengl, Dielectric Spectroscopy in Time and Frequency Domain for HV Power Equipment, ISH Int. Symposium on High Voltage Engineering, Bangalore India, 2001 [2] S.M. Gubanski, P. Boss, G. Cspes, V. Der Houhanessian, J. Filippini, P. Guuinic, U. Gfvert, V. Karius, J. Lapworth, G. Ur- bani, P. Werelius and W. Zaengl, Dielectric Response Methods for Diagnostics of Power Transformers, CIGR TF 15.01.09, Electra No. 202, June 2002 [3] V. Der Houhanessian, Measurement and Analysis of Dielectric Response in Oil -Paper Insulation Syst ems, PhD Thesis, ETH Zuerich, Switzerland, 1998 [4] A. Kachler, T. Leibfried, A. Kchler, B. Breitenbauch, W.S. Zaengl and V. Der Houhanessian, Ageing and moisture analy- sis of power transformer insulation systems, CIGR 2002 ses- sion, paper 12-101, Paris, 2002 [5] B. Breitenbauch, A. Kchler, T. Leibfried and W.S. Zaengl, Insulation Diagnosis by Polarization and Depolarization Cur- rent Measurements, 13th ISH Int. Symp. on High Voltage Engi - neering, Delft, 2003 [6] S.A. Bhumiwat, P. Stattmann, Quality Assurance after Trans - former Refurbishment by means of Polarisation Depolarisation Current Analysis, IEEE-BPT03, Bologna, Italy, 2003 [7] S. Bhumiwat, Insulation Condition Assessment of Transformer Bushings by Means of Polarisation / Depolarisation Current Analysis IEEE Symp. on El . Insulation, Indianapolis, 2004 [8] A. Kchler and T. Bedel , Dielectric Diagnosis of Water Content in Transformer Insulation Systems, ETEP European Transactions on Electrical Power, Vol. 11, No. 1, 2001 [9] L. Pong Review Negative Power Factor Test Results and Case Study Analysis, The 2002 Int. Conference of Doble Clients; Boston/ USA, 2002 [10] A. Kchler, F. Hllmandel, K. Bhm and N. Koch Einfluss uerer Grenzflchen auf die dielektrische Diagnose betriebs- gealterter Hochspannungsdurchfhrungen, ETG Conf. Grenz- flchen in elektrischen Isoliersystemen, Hanau, 2005 [11] A. Kchler, F. Hllmandel, K. Bhm, C. Neumann, N. Koch, K. Loppach, C. Krause and J.J. Alff, Condition Assessment of Aged Transformer Bushing Insulations, CIGR 2006 Session, paper A2-104, Paris, 2006 [12] A. Kchler, F. Hllmandel, M. Appold, M. Liebschner, R. Krump and J. Titze, Diagnose von Hochspannungsdurchfhr- ungen durch PDC-Analyse, ETG-Fachtagung Diagnostik elektrischer Betriebsmittel, Kassel, 19.-20.09.2006 [13] A. Kchler, F. Hllmandel, K. Bhm, N. Koch, P. Brupbacher, and C. Krause, Das dielektrische Verhalten von l-Papier-Iso- lationen unter der Wirkung von Grenzflchen-, Material- und Prfparametern, ETG- Fachtagung Grenzflchen in elektri - schen Isoliersystemen, Hanau, 2005 [14] A. Kchler, F. Hllmandel, K. Bhm, C. Krause and B. Heinrich, Dielektrische Eigenschaften von l-Board- und l- Papier-Isolierungen als Kenngren fr die Diagnose von Trans- formatoren und Durchfhrungen, ETG-Fachtagung Diag- nostik elektrischer Betriebsmittel, Kassel, 2006 [15] A. Kuechler, F. Huellmandel, K. Boehm, M. Li ebschner, Ch. Krause and B. Heinrich, Parameters Determining the Dielectric Properties of Oil Impregnated Pressboard and Presspaper in AC and DC Power Transformer Applications, ISH Int. Symp. on High Voltage Engineering, Ljubljana, 2007 [16] J.J. Alff, V. Der Houhanessian, W.S. Zaengl ans A.J. Kachler; A Novel, Compact Instrument for the Measurement and Evaluation of Relaxation Currents Conceived for On-Site Diag- nosis on Electric Power Apparatus, IEEE Symposium on El. In- sulation, Anaheim, USA, 2000 [17] M. Appold, Ableitung von Diagnosestrategien aus dem dielek- trischen Verhalten von Durchfhrungsisolationen Master Re - search Thesis, FHWS Schweinfurt, 2006