Determination of Water Content in Transformer Solid Insulation by Frequency Domain Spectros
Determination of Water Content in Transformer Solid Insulation by Frequency Domain Spectros
Determination of Water Content in Transformer Solid Insulation by Frequency Domain Spectros
on Electric Power Systems, High Voltages, Electric Machines, Tenerife, Spain, December 16-18, 2005 (pp18-23)
Abstract: - Knowing the water content in transformer solid insulation is basic to determine the transformer
maximum admissible load. As direct measurement of this variable is not possible, indirect techniques should be
used as those based in the measurement of transformer dielectric response in time or frequency domain. By the
application of these techniques is possible to get more accurate information that from traditional methods but
some problems remain related to the interpretation of measurements. This paper is focused in the application of
frequency domain spectroscopy (FDS) to the determination of water content into the different parts of
transformer solid insulation. FDS field measurements over real transformer are reported extracting some general
conclusions that may be useful to perform this kind of measurements.
Key-Words: - Power transformer, insulation diagnosis, dielectric response, FDS, moisture content, field
measurements.
Polarization current is measured after the in the time domain, being so the obtained results are
application of a voltage step to the object under study more reliable.
during the so called charging time. This charging
time should be long enough to allow the polarization 4 FDS measurement and interpretation
processes in the material to be completed. After this in power transformers
time, the material is short circuited and
Power transformer insulation is composed by
depolarization current over the object should be
different dielectric materials (oil, paper, pressboard,
registered.
wood...). Under the effect of an electric field the
The main disadvantages of this method are its
different materials are polarized. The differences in
high sensitivity to electromagnetical disturbances and
the response (in time or frequency) of these materials
the long time required for its application (this
allow separating the effects caused by paper from
disadvantage is common to all the techniques based
those caused by oil. Moreover the dielectric response
in the analysis of the dielectric response) [3].
of the different materials is strongly dependent on
their water content. The analysis of this dependence
3.2 RVM is the basis to get information about the water content
To perform RVM measurements a step voltage is in the materials by means of dielectric response
applied to the object under study during certain measurements.
charging time. The object is then short-circuited There are several factors that influence the
during a short period of time (decharging time) and dielectric response of transformer insulation, e.g. the
finally it is left in open circuit. In this case a voltage geometry of the duct, the dielectric response of oil
appears between the terminals of the material. This (its conductivity and permittivity), the dielectric
voltage is the so called recovery voltage and is response of pressboard or paper, and the temperature
caused by the unfinished depolarization processes in of the whole insulation system. To obtain the
the material. moisture content in solid insulation these variables
The measurement of the recovery voltage should should be known or estimated as parameters during
be repeated for different charging times, obtaining the modeling. For modeling purposes the transformer
the so called polarization spectra. To get the main insulation, consisting of cylindrical pressboard
polarization spectra the same relation between the barriers in series with oil ducts and spacers (Fig. 1), is
charging and decharging time (usually tc=2td) must represented as shown in Fig. 2, where X and Y are
be kept in all the measurements. defined as:
By comparison of polarization spectra with
laboratory curves, obtained over controlled moisture radial thickness of total barriers
samples, an estimation of the moisture content in the X =
object can be extracted. radial width of the duct
In the last years RVM method has been quite total width of the spacers along periphery of the duct
criticized because the water contents estimated by its Y =
periphery of the duct
application are not coherent with that obtained by
direct determination. According to CIGRÉ TF
15.01.09 the problem lays in the simplistic
interpretation of the measurements. Alternatives
have been proposed to that interpretation [2, 4].
3.3 FDS.
FDS method consists in making discrete
measurements of the specimen test impedance in a
frequency range between 1000 Hz and 0.001 Hz.
For this propose, a sinusoidal voltage of variable
frequency is applied to the object under test and the
current flowing across the insulation is registered.
Once the object impedance is known, other relevant Fig. 1. Section of main insulation
parameters may be determined as capacitance (C'), in a core type transformer.
dielectric loss (C'') or tangent δ.
The techniques based in the analysis of dielectric
response in frequency domain (FDS) present less
influence from electromagnetic noise that those based
Proc. of the 5th WSEAS/IASME Int. Conf. on Electric Power Systems, High Voltages, Electric Machines, Tenerife, Spain, December 16-18, 2005 (pp18-23)
shown in Table 1.
C´tap 21
PART OF THE INSULATION WC % C´´tap 21
-9 C´tap 1
Insulation between HV and LV (CHL) 0,3 % 10
C´´tap 1
C´and C´´ (F)
-8
10
As can be seen the registered capacitances in the
three cases are almost the same both in real and in
imaginary part. Just some non significant 10
-9
C´rainy
discrepancies appears in the range of 0,1 Hz C´´rainy
Frequency (Hz)
of creep currents over them.
Fig.2 shows the comparison between
measurements taken with and without rain. As can be Figure 5: CH measurements with and without rain
observed just little differences can be found both in
real as in imaginary part. This results logical if we 5.2 Case 2
take into account that in CHL measurements ground The second case analyzed is an autotransformer
is guarded, so a contribution of currents along the (144/86 kV, 80 MVA) manufactured in 1967.
bushings surface may not appear. In the case of autotransformers, as primary and
10
-8 secondary winding are the same, just capacitance
between HV winding and ground (CH), capacitance
C´rainy between tertiary winding and ground (CT) and
C´´rainy
C´dry capacitance between HV and tertiary winding can be
10
-9 C´´dry determined. .
C´and C´´ (F)
10
-9 water content in that insulation and the estimation in
C´ the other parts is explained.
C´´
-10
10
C´and C´´ (F)
-11
10
-12
10
-13
10
-14
10 -3 -2 -1 0 1 2 3
10 10 10 10 10 10 10
Frequency (Hz)
-7
10
C´
C´´
. 10
-8
C´and C´´ (F)
Frequency (Hz)
some frequencies were negative, that makes no
physical sense.
The low values of the capacitances may be Figure 9: CT measurement
explained by the presence of electrostatic screens
connected to ground between HV and tertiary 6. Conclusion
winding that is typical in autotransformers. In this paper general principles of dielectric
As can be seen in Fig. 8 the major disturbances in response measurements have been described and its
CHT appear in the low frequency region, that is the application to determine moisture content in
part of the curve that provides information about the transformer insulation has been analyzed.
paper condition. In consequence the estimation of Time and frequency domain techniques have
water obtained by MODS for this measurement is not been briefly described making more emphasis in
reliable, and the difference between the estimation of Frequency Domain Spectroscopy (FDS).
Proc. of the 5th WSEAS/IASME Int. Conf. on Electric Power Systems, High Voltages, Electric Machines, Tenerife, Spain, December 16-18, 2005 (pp18-23)
FDS field measurements performed over two Gafvert, V. Karius, J. Lapworth, G. Urbani, P.
power transformers have been reported. The Werelius and W.S. Zaengl, "Dielectric response
following general conclusions can be extracted from methods for diagnostics of power transformers”,
those measurements that can be of general application CIGRE Technical Brochure, No. 254, Paris 2004.
when measuring transformers in field. [3]T.K. Saha "Review of Time Domain Polarization
Measurements for Assessing Insulation Condition
- XY model was conceived to represent the insulation in aged transformers”, IEEE Transactions on
between HV and LV windings. From that, the Power Delivery,Vol 18, No.4, October 2003.
information about the water content in transformer [4] J.P van Bolhuis, E. Gulski, J.J. Smit D. Pepper
insulation that may be extracted from CHL "Interpretation of RVM Measurements, Beyond
measurements is more reliable that those extracted the Polarization Spectrum", Conference Record of
from CH and CL. the 2002 IEEE International Symposium on
- Measurements were compared selecting different Electrical Insulation,Boston MA USA. April 7-10
positions of the LTC, concluding that FDS 2002.
measurements are not sensitive to the LTC selected [5] Insulation Diagnostics Spectrometer IDA,
tap. Programma Electric AB, Eldarv. 4, SE-187 75
- Water or dirt on bushings surfaces can give rise to Täby, Sweden.
creep current along them. Creep currents, in case of
existing, will not affect CHL measurements if guard
cable is properly connected to ground. In the case of
CH and CL measurements is not possible to eliminate
the effect of creep currents and they will be added to
those circulating between HV winding (or LV in case
of CL measurements) and ground causing an increase
in measured losses (C'').
- Important problems appear when measuring very
low capacitances bellow 100 pF.
- The presence of high voltage lines in service close
to the transformer can cause electromagnetic
disturbances that affect the measurements. These
effects are dramatic in case of measuring very low
capacitances.
Acknowledgment:
This work has been carried out at the Electrical
Engineering Department of the University Carlos III
of Madrid, and has been founded by the Spanish
utility Unión Fenosa and by the Spanish Ministry of
Science and Technology (MCYT) under the project
nº DPI2002-09039-C01-02 of the "Plan Nacional de
Investigación Científica, Desarrollo e Innovación
Tecnológica". The authors would like to thank to
Union Fenosa for their interest and support of the
work.
References: