Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Introduction
The importance of moisture management in power transformer is well recognized. Traditionally, the
dryness of transformers in service has been assessed by periodic measurements of the water content in the
insulating oil. This information can be used to assess the moisture content of insulating paper. Excessive
moisture in the solid and liquid insulation can significantly reduce the dielectric strength and partial
discharge inception level. The effect of moisture on insulation aging has been well documented. It has also
been demonstrated that at high temperatures, the residual moisture in winding insulation can trigger the
release of free gas bubbles, thus creating an immediate threat to the dielectric integrity of the insulation
structure. These factors explain well the persistent effort to improve the determination of the moisture
content in the transformer solid insulation.
Moisture in Oil
The traditional method of moisture monitoring calls for oil sampling at regular intervals. The oil sample is
usually processed using a Karl Fischer titration method that provides the total water content in the oil in
ppm. Alternatively, the moisture content can be assessed as relative content in regard to the saturation
value.
One problem with periodic sampling is that the moisture content can vary significantly with the load and
the resulting winding temperature. This is illustrated in Figure 1 where the relative moisture content of a
25-MVA transformer is shown before and after removal from service. This transformer is an old, water-
cooled generating station unit.
100
90
80
70
Oil Temperature, RH%
60
50
40
30
20
10
0
0 5 10 15 20
Days
Figure 1 - Relative Moisture Content Before and After Removal From Service
The measured relative saturation is closely related to the oil temperature in the sensor. At full load, the oil
is in circulation and it can be assumed to be homogeneous. This assumption is valid in normal operation
because the oil circulation is much faster than moisture transfer, so the absolute water content of the oil (in
ppm) can be assumed to be the same through the circulating oil.
Therefore the relative saturation can be determined for any oil temperature within the transformer. The oil
saturation characteristics are shown in Figure 2. This relation is characteristic of Voltesso 35 but the slope
is practically the same for any oil. On this graph, the x axis is 1000/T and the y axis is the log of the
moisture content. On such a graph, the oil relative humidity versus temperature is a straight line. If the oil
characteristics are not known, the line can be traced from two measurements of the same oil at two different
temperatures. When the moisture content of the oil changes, the characteristic line is displaced but remains
parallel to the original one. Thus when the oil is characterized by a specific slope, a single measurement
suffices.
100 100
Relative Moisture Saturation (%)
10 10
5 5
Water-in-oil
condensation: 41 °C
1 1
-40 -20 0 20 40 60 80 100 120
Temperature (°C)
From this graph, it is possible to see at once the temperature that will lead to water condensation in case of
sudden cooling of the transformer under cold ambient temperature. In the example shown in Figure 1 for a
wet transformer, the extrapolated value for 100 % saturation is 41 °C.
From this graph it is also possible to find the relative humidity at any oil temperature. In the transformer
referred to in Figure 1, the hot-spot temperature at full load can be evaluated at 85 °C. It can be assumed
that the oil in contact with insulating paper is at the same temperature as the surface paper and in moisture
14
0 °C
Moisture content of paper (% by dry weight)
12
10
20 °C
8
40 °C
6
60 °C
80 °C
4
120 °C
2
0
0 10 20 30 40 50 60
Relative oil saturation (%)
Figure 3 - Equilibrium Curve for Moisture Partition Between Oil and Paper
Moisture partition curves can be used to determine the moisture content of the paper from the moisture
content of the oil. However these relations can be applied only when the transformer has been under steady
temperature conditions for some time. Such conditions are never truly met on a transformer in service but
quasi-steady conditions can be achieved. To this effect, a method developed by Sokolov(2) can be used to
determine approximately the amount of water in the main insulation. It is reported that if a transformer is
operating at a top-oil temperature between 60 and 70 °C for a period of three days, the rate of moisture
transfer is low enough to indicate that a state close to equilibrium has been reached. The relative saturation
of water in oil is then used to calculate the moisture content of the solid insulation.
Considering the complexity of the adsorption/desorption phenomenon and the complex structure of
transformer insulation, it is not to be expected that the dynamic behavior of moisture migration can be
modelized in a practical way for on-line monitoring. The solution of moisture migration monitoring is to be
attempted by making use of periods of quasi-stability occurring naturally and by introducing some
simplifications to the problem.
Fortunately the insulation of most interest is the turn insulation within the windings. This insulation is made
of thin material and well irrigated by the insulating oil. It is also the insulation component that will be
affected by thermal aging and this is where the bubbling may occur under high overload conditions.
Therefore it is acceptable that calculations of moisture in paper be limited to the paper in the hottest part of
the winding.
The risk of free gas bubbles being released is also a concern limited to high-temperature conditions. It has
been shown that at high temperatures, the residual moisture in the insulation paper can lead to the release of
free gas bubbles. This situation is to be avoided at any rate since it constitutes a major threat to insulation
integrity. Gas bubbles occurring in highly-stressed areas can lead to dielectric breakdown of the main
insulation. A simplified relation between moisture content and bubble inception temperature is shown in
Figure 4. These investigations led by McNutt (7) and later confirmed by Oommen(8) and Davydov (9) indicate
that for a “wet” transformer with a 5 % moisture content in paper, the bubbling temperature is around
100 °C, while for a dry transformer with 0.5 % moisture the bubbling temperature is above 180 °C.
Therefore it is reasonable to focus on accuracy at high temperatures and to accept a larger margin of error at
low temperatures where the moisture content of paper has no immediate consequences.
180
Davydov et al.(9)
Bubble inception temperature (°C) 160
140
120
100
80
Oommen et al.(8)
60
40
0 1 2 3 4 5 6 7 8
Moisture content in paper (%)
Until further experimental work sheds additional light on the diffusion time constant of real transformer
windings insulation, we can use the approximate function shown on Figure 3 as a dotted line. This is the
function used in the FARADAY™ TMCS™ to characterize the diffusion time constant of winding
insulation in the hot-spot area. It can be noted that experimental references always refer to the desorption
phase (paper moisture being released in the oil). For the cooling-down period the absorption process is
known to be much longer, but insufficient data is available to assess the applicable time constant. For
cooling-down periods, utilization of a shorter time constant is a conservative approach since it assumes that
the winding insulation is picking up water at a rate faster than in reality.
Du et al.(12) (theoretical)
80 Von Guggenberg(13) (theoretical)
(14)
Sokolov et al. (theoretical)
60 FARADAY™ Model approximation
40
20
0
1 10 100 1000 10000
Figure 5 - Theoretical and Experimental Diffusion Time Constant on Thin Insulation Material
The model is executed in the sequence outlined in Figure 6. The hot-spot temperature is computed from the
oil temperature, load current and transformer characteristics. If the hot-spot temperature is below 80 °C, no
further calculations are made and the moisture content in paper is assumed to be unchanged from the last
calculated value. If the hot-spot temperature is equal to or above 80 °C, the following parameters are
averaged over the stability period:
• Average hot-spot temperature
• Average moisture content in oil
• Average sensor temperature
From these values is computed the relative humidity of the oil in contact with winding insulation at the hot-
spot temperature. Oil-paper partition curves are then applied to convert the relative humidity in oil to
moisture content in solid insulation. The calculated value of moisture content in the insulation paper is used
to update a register where a default value is kept. This default value is to be used if the hot-spot temperature
and thermal stability do not allow for application of equilibrium curves.
% R.H. °C
Moisture-in-Paper Model
Calculate average values for the computation period
• Moisture content in oil
• Moisture sensor temperature
• Hot-spot temperature
Calculate moisture content in oil at the hot-spot temperature
Calculate moisture content in paper
Conclusions
The water content in transformer oil can be monitored on-line with a relative-humidity sensor such as GE
Syprotec’s AQUAOIL™ 300. This information leads easily to the water condensation temperature. The
relative humidity of the oil in contact with the winding insulation can also be inferred from the relative
saturation characteristics.
In order to assess the moisture content of the solid insulation, some compromises need to be made to deal
with the difficult dynamic pattern of moisture migration between paper and oil. It appears reasonable to
focus on the thin winding insulation paper and disregard the bulk insulation where most of the water is
hiding. The main concerns with moisture in paper are accelerated aging and risk of bubbling. These
detrimental effects are to be feared only in the winding insulation and only at high temperatures. It is
therefore sufficient to limit the accurate assessment of moisture content to the condition of high hot-spot
temperature.
References
1. T.V. Oommen
“Moisture Equilibrium in Paper-Oil Insulation Systems”
Proc. Electrical Insulation Conference, Chicago, October 1983
2. V. Sokolov and B. Vanin
“In-Service Assessment of Water Content in Power Transformers”
Doble Client Conference, Boston, May 1995
3. T.V. Oommen
“On-Line Moisture Sensing in Transformers”
Proceeding of the 20th Electrical/Electronic Insulation Conference, Boston, October 1991
4. H. Azizian, J.A. Proskurnicki and J.G. Lackey
“Relative Saturation Versus Moisture Content of Insulating Oil and its Application in Monitoring
Electrical Equipment”
Doble Client Conference, Boston, 1995
5. V.G. Davydov, O.M. Roizman and W.J. Bonwick
“Evaluation of Water Content in Transformer Insulation Systems”
EPRI Substation Equipment Diagnostic Conference VI, New Orleans, February 1998
6. IEEE Standard C57.91-1995
“IEEE Guide for Loading Mineral-Oil-Immersed Transformers”
7. W.J. McNutt, G.H. Kaufmann, A.P. Vitols and J.D. MacDonald
“Short-Time Failure Mode Considerations Associated With Power Transformer Overloading”
IEEE Trans. PAS, Vol. PAS-99, No. 3, May/June 1980
8. T.V. Oommen, E.M. Petrie and S.R. Lindgren
“Bubble Generation in Transformer Windings Under Overload Conditions”
Doble Client Conference, Boston, 1995
9. V.G. Davydov, O.M. Roizman andW.J. Bonwick
“Transformer Insulation Behavior During Overload”
EPRI Substation Equipment Diagnostic Conference V, New Orleans, February 1997
10. B. Noirhomme, B. Sparling, J. Aubin and P. Gervais
“A Practical Method for the Continuous Monitoring of Water Content in Transformer Solid Insulation”
EPRI Substation Equipment Diagnostic Conference V, New Orleans, February 2000
11. P.J. Griffin
“Water in Transformers – So What!”
Doble Client Conference, Boston, May 1997
12. Y. Du, M. Zahn, B.C. Lesieutre, A.V. Mamishev and S.R. Lindgren
“Moisture Equilibrium in Transformer Paper-Oil Systems”
IEEE Electrical Insulation Magazine, January/February 1999
13. P.A. Von Guggenberg
“Discussion of Sokolov and Vanin Paper” (Ref. 2)
Doble Client Conference, Boston, May 1995
14. V. Sokolov, P. Griffin and B. Vanin
“Moisture Equilibrium and Moisture Migration Within Transformer Insulation System”
Doble Client Conference, Boston, May 1999