On-line Analysis of Power Transformer Bushings

M. Faifer, R. Ottoboni, L. Cristaldi, S. Toscani

Dipartimento di Elettrotecnica, Politecnico di Milano
Piazza Leonardo da Vinci 32 – 20133 Milano ITALY
e-mail: marco.faifer@polimi.it

Abstract — The approach to the maintenance of devices and the other side, efforts are made to innovate and to improve the
systems has strongly changed in the last years, thanks to the reliability of the electrical system itself. In particular there is a
availability of more and more sophisticated technologies. In growing interest about the employment of advanced
particular, when the servicing involves very critical processes, a maintenance strategies such as CBM (Condition Based
new interesting strategy is represented by the Condition Based Maintenance): the service activity is no longer based on the
Maintenance (CBM). It is based on the continuous observation of expected lifespan of the components, but it is planned starting
some physical parameters which permits to infer the status of the from the continuous monitoring of their status.
system and thus to detect the need to service it. In this paper a
novel monitoring method is proposed. It allows the application of The aim of this paper is to propose a technique allowing the
CBM techniques to one of the least reliable components in HV implementation of CBM for high voltage systems and in
and MV systems: the transformer bushing. particular for one of their least reliable component: the
Starting from some theoretical and practical considerations, it transformer bushing. It has been estimated that they are
will be shown that it is possible to measure the capacitance and responsible for almost half of the failures occurring in High
the power factor of transformer bushing, without putting the Voltage (HV) and Medium Voltage (MV) transformers [2].
plant out of order. It is known that these two parameters are
particularly useful to schedule the maintenance activity. Until now, the status of the bushings has been assessed
This goal can be achieved by superimposing a high frequency through complex procedures which usually require to put the
sinusoidal voltage between the bushing tap and the ground. Some transformer offline and sometimes even to remove it. The
considerations prove that having chosen a proper frequency, the limits of this approach are clear: the status of the bushing
consequent current is much more sensitive to the bushing cannot be assessed continuously and the cost is extremely high.
capacitance than to the other impedances. Finally the Moreover the electrical utilities are prone to reduce the
architecture of a measurement system is also presented. frequency of the tests: this unavoidably raises the fault
probability.
Keywords- diagnostic, power transformer bushings, power Basically, these are the reasons which pushed to investigate
factor and capacitance. for new methods allowing to continuously assess the wear
status of transformer bushings. In the literature some papers
I. INTRODUCTION reporting on-line diagnostic method can be found [2-5]. All of
It is a matter of fact that the importance of the electrical them perform a dynamic analysis of diagnostic indexes that are
energy is continuously growing nowadays. Everyone is aware someway related to the main characteristic parameters of the
of the increasing number of electrical, electronic and automatic bushings [3]: capacitance and power factor. The assessment of
devices both in industrial processes and in everyday life. the status of the bushings is based on the time evolution of the
Moreover, it is clear that the new infrastructures and services parameters (or indexes) rather than on their instantaneous
are designed to achieve maximum efficiency and comfort, values.
objectives that are usually met through a massive employment According to the aforementioned approach, in this paper
of electrical-powered systems. These are so widespread that a the theoretical background of a new method for the assessment
sudden interruption of their operation often results in huge of the wear status of power transformer bushings are presented.
discomforts and sometimes could dramatically affect the The technique has been developed with the aim to allow the
regular carrying out of human activities. Meaningful examples on-line monitoring of the device and to overcome some typical
are the blackouts occurred in the last years: many city districts limits of the other methods presented in the literature.
and industrial areas were put offline and almost all sort of
activities were forced to stop (Italy September 2003, USA and
Canada August 2006, Italy, Germany and France November
2006, etc.). II. STATE OF THE ART
The first method reported in the literature we consider is the
These facts show the increasing importance of an
so called “sum current method” [4]. This approach is based on
uninterrupted power supply. The grid operators are focusing
the Kirchhoff current law, and it analyses the sum of the three
their efforts to improve the reliability of the transmission and
currents drawn by known impedances connected to the bushing
distribution systems through a tighter coordination between
taps. Supposing that the line voltages are symmetric and both
them and the adjustment of the management policies [1]. On

978-1-4244-7935-1/11/$26.00 ©2011 IEEE

the bushing impedances and the known impedances are operating at a frequency in the order of some megahertz. In the
balanced, the zero sequence current is null. Any variation of a following it will be proved that, in this way, it is possible to
bushing impedance results in a change of the homopolar evaluate just the impedance of the bushing, minimizing the
current. Therefore it is possible to infer information about the influence of the transformer and the substation devices on the
wear status of the bushing through the observation of the zero measurement.
sequence current. Unfortunately this analysis is not simple and
Therefore, in order to evaluate the feasibility and the
in many cases it requires the use of complex tools such as
performance of the proposed method, some considerations
neural networks.
have to be done about the model of the whole substation in the
The method allows achieving good results but, in our frequency range taken into account. In particular four sections
opinion it suffers from several limitations: have been identified: the bushing, the power transformer, the
lines and the other substation equipment.
• it supposes that the power network is three phase
symmetric; A. Bushing
• it does not allow the direct analysis of a single For a first analysis a particular type of bushing has been
bushing; considered: oil impregnate paper (OIP) bushings. The typical
• it does not take into account multiple failures; model of a bushing at the AC mains frequency is made-up by
• the training of the neural network is time consuming two capacitor, C1 and C2 with two parallel resistances R1 and R2
and requires the availability of a fairly high number representing the power losses [6]. C1 is the equivalent
of failure cases. capacitance between the bushing tap and the high voltage
These limits have been partially overcome in the methods terminal, while C2 is the equivalent capacitance between the
presented in [2] and [5]. The basic idea is to measure also the bushing tap and ground. When considering the AC mains
line voltages, in order to compensate the effect of the voltage frequency, typical values of the bushing parameters are some
unbalance. hundreds of picofarads for the capacitances and few gigaohms
for the resistances. At higher frequencies it can be shown that
The method presented in [2] requires voltage sensors as the capacitance values are basically the same, while the
external references in order to improve the performances of the resistance values change considerably [9], [10] and [11]. In
“sum current method”, but it is still affected by the problems particular at 1 MHz the resistances are some tens of kiloohms.
related to the employment of neural networks. In [5] the In the same papers it is also shown that both the capacitance
reference is provided by a voltage transformer. This introduces and the resistance values depend on the moisture concentration
criticalities in terms of reliability and cost. In fact for this into the capacitor even at frequencies of some megahertz.
method it is essential to guarantee an excellent stability of the
characteristics of the employed devices. In the bushing model shown in Figure 1 additional
parameters (Rs and Ls) have been introduced. These represent
the effects of the central conductor of the bushing that connects
the line with the transformer [7]. Typical values are: Rs some
III. METHOD AND POWER SYSTEM MODELS tenths of ohms, Ls some microhenries.
The basic idea of the proposed method is to perform a
direct evaluation of the capacitance and power factor (PF) of
the bushing by means of voltage and current measurements
performed at its tap. Through the superimposition of a proper
voltage signal at the tap, the measurement of the consequent
current and the vector analysis of the acquired data, it is
possible to estimate the bushing capacitance and PF. These two
parameters are recognized as good indicators of the wear status
of the bushing [6] [8]. For example, they allow to identify one
of the most frequent problems in the bushing: the dielectric
degradation due to moisture infiltrations.
From a theoretical point of view, the measurement of
Figure 1. Bushing model
voltage and current at the bushing tap allows to evaluate the
equivalent impedance across this port. This impedance is due
B. Power transformer
to the combined effect of the proper impedance of the bushing
together with those of the other devices connected to it. Since Many high-frequency models of power transformers can be
the final target is the evaluation of the parameters of the found in literature [12] [13] [14]. They basically share the same
bushing, it is mandatory to evaluate how the presence of the circuit topology. For this analysis that presented in [14] has
other devices affects the impedance measured at the tap and if been taken into account: it describes the power transformer
it is possible to operate in order to minimize their effects. As through an electrical network which is simple yet detailed
far as this last problem is concerned, an efficient solution is enough (Figure 2). This model has been proposed for Power
represented by performing impedance measurement at a Line Carrier (PLC) applications; measurements confirm that at
frequency much higher than that of the AC mains. In particular, high frequency the iron core does not affect the impedance
the performed analysis shows that good results can be obtained across the terminals. Through the analysis of the transformer
circuit at frequencies of some megahertz, it can be shown that the impedance measured between the tap and ground is
the input impedance between HV or LV terminals and ground essentially due to the bushing.
is essentially capacitive and weakly depends on the impedances
On the contrary, when the line is connected, it is not
connected to the others nodes.
possible to a priori define the equivalent impedance at the HV
terminal of the bushing since the impedance due to the
transmission line is unpredictable. For instance, when its
behavior becomes inductive, it is possible to have resonance
effects due to the capacitances of to the other devices, which
does not permit to properly evaluate the parameters of the
bushing. A possible solution to this problem is to install proper
filters similar to those usually employed for power line carrier
applications. In fact, the insertion of this kind of filters, called
“line traps” or “line tuners”, allows decoupling the line from
the substation or better, to introduce a very low impedance
between the line and ground at the considered frequency.
In any case, in the high frequency equivalent circuit there is
an impedance connected between the high voltage terminal of
the bushing and the ground, whose value affects the accuracy
of the evaluation of the bushing parameters.

IV. METHOD FEASIBILITY AND ACCURACY

Figure 2. Power transformer model. Starting from the previous considerations, in this section an
evaluation of the feasibility and the theoretical performance of
the proposed method is presented. As aforementioned, the aim
C. Lines and substation equipment of the method is to evaluate the status of the bushing by
measuring the value of the equivalent capacitance given by the
The power lines connected to the substation cannot be parallel of C1 and C2. For this analysis a simplified model of the
modeled by means of concentrated parameter models. In fact, system has been introduced. In particular it has been supposed
because of the line lengths and the wavelengths involved at the that the line is disconnected or that there is a filter that allows
considered frequencies, the phase delay and the reflections decoupling the power line. As stated before, in this case the
cannot be neglected [15]. impedance connected between the high voltage terminal of the
On the contrary, a concentrated parameter model can be bushing and ground can be basically represented by a single
defined for great part of the substation equipment. In fact, the equivalent capacitance Cload (Figure 3).
involved dimensions are significantly lower than a quarter of
the wavelength at 1 MHz (approximately 75 m). In particular
voltage measurement transformers, current measurement
transformers, surge suppressors, switchgears and insulators can
be represented by means of equivalent capacitances connected
to ground.

D. Consideration on the overall model

By considering the overall model made up of the four
sections previously defined, two particular conditions can be Figure 3: equivalent model of the system.
defined: the power line on the same side of the bushing can be
connected or disconnected. This is important because it affects In order to roughly understand the dependency of the
the possibility to estimate the equivalent impedance connected capacitance measured at the tap Ctap by Cload, let us neglect the
to the high voltage terminal of the bushing. In fact, as stated effect of the resistances R1 and R2. Furthermore, let us assume
before, the transformer and the substation equipment effects that the impedance due to the series of Rs and Ls is much lower
can be represented by simple capacitances at the considered than that due to Cload at the measurement frequency. Under
frequency. As a consequence, if the line is disconnected, the these conditions, the equivalent capacitance measured at the tap
equivalent impedance between ground and the high voltage is:
terminal of the bushing can be reduced to a capacitance. With ‫ܥ‬ଵ ȉ ‫ܥ‬௟௢௔ௗ
regard to Figure 1, this means to have two capacitances ‫ܥ‬௧௔௣ ൌ ‫ܥ‬ଶ ൅ (1)
‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ
connecting the terminals “Line” and “Transf” to ground. If
their impedances at the considered frequencies are low enough,
 The following sensitivity coefficients of (1) can be In other terms:
introduced: ଵ
ο‫ܥ‬ଶ ൅ ο‫ܥ‬ଵ మ
ଶ ಴భ
߲‫ܥ‬௧௔௣ ‫ܥ‬௟௢௔ௗ ͳ ൬
಴೗೚ೌ೏
ାଵ൰
(9)
ൌ ൌ ଶ (2) ൒݇
߲‫ܥ‬ଵ ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻଶ ஼భ ଵ
ȉ ο‫ܥ‬௟௢௔ௗ
ቀ ൅ ͳቁ ಴ మ
஼೗೚ೌ೏ ቀ ೗೚ೌ೏ ାଵቁ
಴భ
߲‫ܥ‬௧௔௣ ‫ܥ‬ଵ ଶ ͳ (3)
ൌ ൌ ଶ with an arbitrary constant k much greater than 1.
߲‫ܥ‬௟௢௔ௗ ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻଶ ஼೗೚ೌ೏
ቀ ൅ ͳቁ
஼భ (9) is definitely satisfied if both of the following conditions are
߲‫ܥ‬௧௔௣ (4) satisfied:
ൌͳ
߲‫ܥ‬ଶ ݇ ͳ
ଶ ȉ ο‫ܥ‬௟௢௔ௗ ൑ ο‫ܥ‬ଵ ȉ ଶ
These equations clearly show that the higher is the ratio ቀ
஼೗೚ೌ೏
൅ ͳቁ ቀ
஼భ
൅ ͳቁ
Cload/C1 the lower is the influence of Cload on Ctap. Therefore, in ஼భ ஼೗೚ೌ೏
order to obtain a measurement at the tap depending only by the
parameters of interest (C1 and C2) it is necessary to guarantee a ݇ (10)
ȉ ο‫ܥ‬௟௢௔ௗ ൑ ο‫ܥ‬ଶ
proper value of Cload. In particular, if Cload >> C1, Ctap ≅ C1//C2. ஼೗೚ೌ೏ ଶ
The error due to Cload in the evaluation of the equivalent ቀ ൅ ͳቁ
஼భ
capacitance C1//C2 is:
‫ܥ‬ଵ ȉ ‫ܥ‬௟௢௔ௗ
ߝ ൌ ൬‫ܥ‬ଶ ൅ ൰ െ ሺ‫ܥ‬ଵ ൅ ‫ܥ‬ଶ ሻ (5)
‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ Typically C1 ” C2, hence it follows that the first condition of
Having fixed a maximum allowed error E, it can be (10) is the most critical.
established a condition on Cload:
Therefore in order to obtain a measurement of ¨Ctap weakly
‫ܥ‬ଵ ଶ  ‫ܥ‬ଵଶ െ ‫ܥܧ‬ଵ affected by ΔCload the following equation has to be satisfied:
ቤ ቤ ൏ ‫ ܧ‬ሱۛۛۛሮ ‫ܥ‬௟௢௔ௗ ൐ (6)
ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻ ‫ܧ‬ ‫ܥ‬௟௢௔ௗ ଶ
݇ ȉ ο‫ܥ‬௟௢௔ௗ ൑ ൬ ൰ ȉ ο‫ܥ‬ଵ (11)
By supposing a capacitance C1=100 pF and a maximum ‫ܥ‬ଵ
error E =1 pF (it corresponds to 1% of C1, that is the
capacitance variation taken into account by IEEE standard [3]) (11) can be written in term of per-unit values as:
the limit given by (6) on Cload is 9.9 nF.
ο‫ܥ‬௟௢௔ௗ ‫ܥ‬௟௢௔ௗ ͳ ο‫ܥ‬ଵ
Actually, as stated in the previous section, in measurements ൑൬ ൰ȉ ȉ  (12)
‫ܥ‬௟௢௔ௗ ‫ܥ‬ଵ ݇ ‫ܥ‬ଵ
for diagnostic it is important to evaluate the time evolution of
the parameters more than their absolute values. Therefore it is From (12) becomes that the relative variation ΔCload/Cload
worthy to understand how the variations of the parameters of must be lower than (Cload/C1)/k-times the relative variation
the model cause a variation of Ctap.
ΔC1/C1. If we assume a ratio Cload/C1 equal to 10 (the lower
Taking into account (1), (2), (3) and (4), the variation of limit previously computed in order to satisfy (6)) and we allow
capacitance ¨Ctap can be written as: a maximum error of 10% in the evaluation of Ctap because of
the variation of Cload (k = 10), we can accept a relative variation
ͳ ͳ of Cload equal to that we are required to detect on C1 (1 %).
ο‫ܥ‬௧௔௣ ؆ ο‫ܥ‬ଶ ൅ ଶ ȉ ο‫ܥ‬ଵ ൅ ଶ
ቀ
஼భ
൅ ͳቁ ቀ
஼೗೚ೌ೏
൅ ͳቁ Therefore assuming a relative variation of Cload of the same
஼೗೚ೌ೏ ஼భ (7) order of that of C1, the evaluation of the changes of the bushing
ȉ ο‫ܥ‬௟௢௔ௗ parameters C1 and C2 by means of the measurements of Ctap
can be assured with an error lower than 10%.
Since we are interested in measuring just the variation of In order to confirm the simplifications adopted until now, a
Ctap due to those of C1 and C2, the following condition shall be Monte Carlo simulation of the complete model shown in
verified: Figure 3 has been performed. In particular it has been
ͳ ͳ simulated the changes of the impedance measured at the tap
ଶ ȉ ο‫ܥ‬௟௢௔ௗ ‫ ا‬ο‫ܥ‬ଶ ൅ ଶ ȉ ο‫ܥ‬ଵ (8) (Ctap and Rtap of the parallel model) due to the variation of the
஼೗೚ೌ೏ ஼భ parameters Rs, Ls and Cload. The simulation has been performed
ቀ ൅ ͳቁ ቀ ൅ ͳቁ
஼భ ஼೗೚ೌ೏ by forcing a relative variation of these parameters lower than
10%. The rated values of all the parameters except Cload are the
same for each simulation. On the contrary, different rated
values of Cload have been considered: 30 nF e 100 nF. The other
parameters of the simulations are: Rs = 0.2 Ω, Ls=1 μH, C1 = C2
= 300 pF, R1 = R2 = 15kΩ.
 The simulation results reported in Table 1 confirm the C1: the relative variation of C2 causes a relative variation of Ctap
previous simplified analysis. In particular in the last column the about 2 times smaller. Therefore, supposing that we have to
three times the p.u. standard deviation is reported. It is clear detect variations of the two capacitances greater than 1%, the
that in the complete model of Figure 3 the absolute error in measurement system must feature a resolution better than
evaluating the parameter of the bushing is affected by the 0.1%.
presence of the resistance Rs and the inductance Ls. Because of
their typical values, the last one has the greater influence. Similar considerations apply to the evaluation of the
However the values of this inductance can be supposed to be resistances R1 and R2 even if the standards do not provide a
lower enough to satisfy the previously discussed condition maximum variation of these parameters but just a threshold. As
explained in (6). It can be computed that an inductance of reported in the previous section the values of these parameters
are very dependent from the frequency. Therefore, new
about 1 μH, having replaced Cload with a short circuit, causes an
thresholds should be defined at the considered measurement
error of 0.6% and 2.2% in the evaluation of the equivalent
frequency. Anyway, it is reasonable to consider that the relative
capacitance and resistance of the bushing respectively.
variations are about the same. From these considerations, an
Furthermore, Ls mainly depends on the geometry of the system
adequate resolution of the measurement system for the
and hence it can be considered very stable.
estimation of Rtap can be some percent.
Table 1: Monte Carlo simulation results (1000 iterations).
It is clear that the same requirements for the measurement
system are even more valid when the high voltage terminal of
Relative
Cload Mean Standard percentage the bushing is connected to ground by means of a proper LC
Parameter resonant filter.
[nF] value deviation deviation
k=3

Ctap [pF] 600.54 0.20 0.10 V. MEASUREMENT SYSTEM

30
Rtap [Ω] 7446.5 5.6 0.22 The proposed method is based on a volt-amperometric
(vector) measurement of both the real and the imaginary part of
Ctap [pF] 602.66 0.21 0.10 the input impedance across the bushing tap. The measurement
100 system should provide the accuracies defined in the previous
Rtap [Ω] 7397.7 5.6 0.22
section as well as assure a proper insulation between the high
voltage system and the user. A possible solution is shown in
Figure 4 where a high frequency transformer is used in order to
A further aspect that has to be investigated is the inject the high frequency test signal and to isolate the signal
measurement accuracy required for obtaining reliable results. generator by the high voltage system.
This evaluation can be easily done by rewriting in per-unit the
equation (7) and taking into account the typical values of the
parameters:
ο‫ܥ‬௧௔௣
‫ܥ‬௧௔௣
ο‫ܥ‬ଶ ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻ ȉ ‫ܥ‬ଶ ο‫ܥ‬௟௢௔ௗ
؆ ȉ ൅
‫ܥ‬ଶ ‫ܥ‬ଵ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଶ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଵ ȉ ‫ܥ‬ଶ ‫ܥ‬௟௢௔ௗ
‫ܥ‬ଵ ଶ ȉ ‫ܥ‬௟௢௔ௗ
ȉ (13)
ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻ ȉ ሺ‫ܥ‬ଵ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଶ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଵ ȉ ‫ܥ‬ଶ ሻ
ο‫ܥ‬ଵ
൅
‫ܥ‬ଵ Figure 4: measurement scheme.
ଶ
‫ܥ‬௟௢௔ௗ ȉ ‫ܥ‬ଵ
ȉ Furthermore, the transformer allows establishing a low
ሺ‫ܥ‬ଵ ൅ ‫ܥ‬௟௢௔ௗ ሻ ȉ ሺ‫ܥ‬ଵ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଶ ȉ ‫ܥ‬௟௢௔ௗ ൅ ‫ܥ‬ଵ ȉ ‫ܥ‬ଶ ሻ impedance conductive path between the tap and ground at the
AC mains frequency (Figure 5).
The typical value of the capacitance C1 is in the range 300
pF - 600 pF while the capacitance C2 may be one to four times This feature is very important since it avoids the presence
greater than C1. Starting from these data it is possible to of a dangerous voltage on the tap because of the line voltage
evaluate the accuracy of the measurement system. The worst and the capacitive coupling. So it is possible to have a very low
case for the evaluation of the variations of C1 occurs when C2 is voltage at the main frequency across the tap allowing the
four times greater than C1. In fact, considering again a ratio connection of the electronic devices necessary for the
Cload/C1 of 10, the relative variation of C1 causes a relative measurement of the voltages V1 and V2. The first one is the
variation of Ctap about 6 times smaller. The worst case for the voltage applied to the tap, while the second one is proportional
evaluation of the variations of C2 happens when C2 is equal to to the current flowing into it. The two voltages are properly
amplified and then quantized by means of analog digital
converters. The data stream is transmitted to a processing unit [3] IEEE Standard General Requirements and Test Procedure for Power
through an optical channel. This solution guarantees a proper Apparatus Bushings, IEEE Standard C57.19.00, 2004.
insulation between the bushing tap and the final user. [4] M.F. Lachman, W. Walter, P.A. von Guggenberg, "On-line diagnostics
of high-voltage bushings and current transformers using the sum current
method," IEEE Trans. on Power Del., vol. 15, no. 1, pp.155-162, Jan.
2000.
[5] Z. Berler, V. Sokolov, V. Prikhodko, D. Bates, "On-line monitoring of
HV bushings and current transformers", in Proc. Electrical Insulation
Conf. and Electrical Manufacturing Expo., Indianapolis, IN, 2005,
pp.61-66.
[6] Insulated bushings for alternating voltages above 1000 V, IEC Standard
60137.
[7] M. Wang, A.J. Vandermaar, K.D. Srivastava, “Improved detection of
power transformer winding movement by extending the FRA high
frequency range,” IEEE Transactions on Power Delivery, vol. 20 , no.3
pp 1930-1938, July 2005
[8] ABB Power Technology Products AB (2000, Aug. 30). Bushing
diagnosis and conditioning - ABB Product information [Online].
Available:
http://library.abb.com/global/scot/scot252,nsf/veritydisplay/f65a6f910c4
51bc1c1256bed00302ee3/$File/2750%20515-
Figure 5: system schematic at the main frequency. 142%20en%20Rev%200.pdf
[9] R. Venkatesh, S.R. Kannan, "Diagnostic testing and condition
The measurement of the impedance can be performed using monitoring of transformer bushings", in Proc. Annu. Report Conf. on
a frequency domain approach. In particular FFT-based Electrical Insulation and Dielectric Phenomena, Cancun, Quintana Roo,
algorithms can be used for this purpose, which allow simple yet Mexico, 2002, pp. 882-886.
very accurate measurements [16]. [10] C. Ekanayake, S.M. Gubanski, A. Graczkowski, K. Walczak,
"Frequency response of oil impregnated pressboard and paper samples
for estimating moisture in transformer insulation", IEEE Trans. on
Power Del., vol. 21, no. 3, pp.1309-1317, July 2006.
CONCLUSION [11] L. E. Lundgaard, D. Linhjell, O.L. Hestad, J.-T. Borlaug, "High
frequency dielectric response of paper/oil insulation", Proc. IEEE Int.
The paper represents a first step towards the introduction of Conf. on Dielectric Liquids, Poitiers, France, 2008, pp.1-4.
CBM techniques to very critical devices such as MV and HV [12] M. Popov, L. van der Sluis, G.C. Paap, "A simplified transformer model
transformer bushings. In particular, it has been presented a for the simulation of fast surges and adequate protection measures", in
method for the measurement of the bushing capacitance and Proc. IEEE Power Engineering Society Winter Meeting, Columbus, OH,
power factor, which just require to connect a device to the 2001, vol. 1, pp. 323-328.
bushing tap. Starting from a high frequency modeling of all the [13] JA. Martinez, R. Walling, B.A. Mork, J. Martin-Arnedo, D. Durbak,
substation equipment, it has been shown how it is possible to "Parameter determination for modeling system transients - Part III:
Transformers", IEEE Trans. on Power Del., vol. 20, no. 3, pp. 2051-
obtain reliable information about the capacitance and power 2062, July 2005.
factor of the bushing, even when the monitored bushing
[14] T. Tran-Anh, P. Auriol, T. Tran-Quoc, "High frequency power
transformer is operating. A sensitivity analysis of the proposed transformer modeling for Power Line Communication applications", in
method is discussed and the architecture of the required Proc. IEEE Power Systems Conf. and Expo, Atlanta, GA, 2006. pp.
measurement system is presented. 1069-1074.
[15] Central Station Engineers of the Westinghouse Electric Corporation,
“Regulation and Losses of Transmission Lines” in Electrical
Transmission and Distribution Reference Book, 4th ed. Pittsburgh:
REFERENCES Westinghouse Electric Corporation, 1950, pp. 265-289.
[1] A. Berizzi, "The Italian 2003 blackout", in Proc. Power Engineering [16] M. Faifer, R. Ottoboni, S. Toscani, L.Ferrara, “Non-destructive Testing
Society General Meeting, Denver, CO, 2004, vol.2, pp.1673-1679. of Steel Fiber Reinforced Concrete using a Magnetic Approach,” IEEE
[2] A. Setayeshmehr, A. Akbari, H. Borsi, E. Gockenbach, "On-line Transactions on Instrumentation and Measurement, to be published
monitoring and diagnoses of power transformer bushings", IEEE Trans.
Dielectr. Electr. Insul., vol. 13, no. 3, pp.608-615, June 2006.