THE EFFECTS ON WINDING CLAMPING PRESSURE DUE TO CHANGES IN

MOISTURE, TEMPERATURE AND INSULATION AGE

By Tom Prevost, EHV-Weidmann, USA
David J. Woodcock, Weidmann Technical Services Inc., USA
Christoph Krause, H. Weidmann Ltd., Switzerland

I.

ABSTRACT

Power transformer windings are designed to withstand high axial forces which
result from short circuit events. To withstand these forces, the winding assembly is
clamped to a predetermined pre-load pressure during manufacture. This paper
discusses the important relationship between changes in moisture level versus
clamping pressure for new transformers. It further relates laboratory investigations of
changes in clamping pre-load versus changes in operating temperature, moisture
and the insulation aging effect.
II.

INTRODUCTION

An important criteria for prolonged transformer life is its ability to withstand

short circuit events. A short circuit can result in severe radial and axial forces which
can damage the insulation integrity. Axial forces which result from short circuits are
controlled by the proper pre-loading of the transformer coils during manufacture. The
windings are pre-loaded to a pressure which is at least as high as the maximum
calculated axial short circuit force.(1) As long as the transformer clamping system
maintains pre-load pressure, the windings will remain tight during a short circuit event
and should therefore not sustain any damage, due to movement of the conductors.
In most transformers designed today, a rigid clamping system is utilized to
compress the windings to the specified pre-load value. Any change in the thickness
of the materials in the winding and associated end insulation will change the pressure
on the winding. The thickness of the conductor material will not change except for the
thermal expansion and contraction during load cycles. The cellulose insulation
material, being organic, will change in thickness and elasticity over time resulting
from the effect of moisture, temperature, and aging.
This paper discusses tests designed to measure the tendency of the cellulose
insulation to expand and shrink during manufacturing processes and through the life
of a transformer. Tests were carried out on stacks of cellulose pressboard spacers
alone and also cellulose pressboard combined with cellulose insulated copper
conductors. Aging tests were carried out on stacks of cellulose pressboard spacers
1

arranged in vessels containing the combination of materials found in typical power

transformers.
Methods are reviewed of those manufacturing processes which are utilized to
ensure tightness of the windings. Test results demonstrate that some key parameters
need to be considered and controlled during the manufacturing process and during
operation.
III.

MANUFACTURING METHODS FOR SHORT CIRCUIT PERFORMANCE

An ideal insulating spacer material in the coil stack would have purely elastic
behavior. Although this is not practically attainable, it can be approached by the use
of machined high density pressboard which has been pre-stressed and stabilized in
the transformer manufacturing process.(2) The manufacturing process should include
the following procedures in order to assure that the windings maintain the proper preload forces:

Drying of the winding must be carried out under pressure

The windings must be sized and stabilized
The windings must be re-tightened after oil impregnation.

Figure 1 demonstrates the change in thickness of transformerboard spacers

during coil processing. The pressure on the spacers was maintained through the use
of springs. Of interest is the additional shrinkage of the insulation material due simply
to the oil impregnation process. This shrinkage has been confirmed by transformer
manufacturers although it is difficult to explain physically. Tests to quantify this
phenomenon are presently in progress.

Variation of Spacer Thickness During Processing

Spacer Stack Height (mm)

112
110
108

<-- 7 Days-->

106
104
102
100
98
96
94
As-Is ( 6% m/c, 1
N/mm^2)

Compressed ( 6 %
m/c, 10 N/mm^2)

Vacuum Dried (10

N/mm^2)

Oil Impregnated (10

N/mm^2)

Effect of coil processing on the thickness of a 100 mm stack of

Transformerboard spacer material. Material was vacuum dried for 7
days at 0.5mbar under a load of 10 N/mm2
FIGURE 1
IV.

EFFECTS OF MOISTURE

Transformer insulation components, by the nature of the cellulose fibres they

are made of, have a tendency to change thickness. This effect is caused mainly by
the presence of water in the system and the variation of the temperature. Therefore, in
the case of rigid clamping distances, the clamping force will vary according to the
tendency of the cellulose to expand and shrink.
Little data can be found in literature for assessing the usually unrecorded, therefore
undetected variation of the static clamping forces. Laboratory scale experiments
were performed with pressboard components stacked in a rigid clamping fixture.( 3)
The first test series (A) was run with dried models, and the second series (B) with
dried, oil impregnated and oil immersed models. In each test series, two different
types of models were investigated: one consisted solely of a stack of radial spacers,
the other of radial spacers interspersed with segments of insulated copper
conductors.
V.

TEST ARRANGEMENT AND PROCEDURE

The pressboard used for all tests was pre-compressed Transformerboard
(Weidmann TIV) with milled surfaces on both sides of the board.(4)
3

Model-1: Radial Spacer Stack

The stack was 32 mm in height, composed of 16 square spacer pieces, 2 mm
thick and 25 mm wide.
Model-2: Spacers with Insulated Copper Conductor Segments
The stack consisted of 5 spacers (2 mm thick), separated by bundles of
copper conductors according to (Figure 2).

copper 12 x 3, R 1 mm
Kraft paper insulation 1 mm
pressboard spacer 2 mm

40

25

25

Model-2: spacers (2 mm thick) interspersed with paper insulated copper

conductor disk segments. Apparent pressed area: 625 mm2.
FIGURE 2

Test Procedure
Series (A): Dried Specimens
Pre-processing: The samples were vacuum (< 0.5 mbar) dried at 105C for 7
days under a constant load of 5 N/mm2. Then, the models were cooled down to
room temperature (still under vacuum).
Test: The samples were placed in the test device and a pressure of 2.5 N/mm2
was applied. The models were exposed to a controlled atmosphere of 23C
and 50 % relative humidity for 24 hours. Subsequently, they were slowly dried
again by applying vacuum at room temperature.
Series (B): Dried and Oil Impregnated Specimens
Pre-processing: The same drying, pressing and assembly procedures apply as
for series (A).
Test run: The samples were placed in a test device and a pressure of 2.5
N/mm2 was applied. The test device was then placed in a vacuum chamber and
the models were oil impregnated under vacuum. They were then kept immersed
in oil in a small container of approx. 300 ml.. These models were exposed to a
controlled, moist atmosphere (23C and 75% relative humidity) for 450 hours.
After this period, the oil was drained and vacuum was applied. In order to
accelerate the drying process, the temperature was increased to 60C, for 100
hours. Finally, the heat was switched off and the models cooled down to room
temperature within less than 24 hours.
Clamping of the test-device and measuring methods
Figure 3 depicts the test-devices for both model types. A steel block was
placed on the models, and the pressure sensor (load cell, nominal compression
< 10 m/kN) was positioned between the steel block and the upper clamping
plate. The stacks were rigidly clamped between two steel plates, and the
clamping force was set by adjusting the nuts. The moisture content for series (A)
with non-oiled samples was measured by recording the weight variation of a
control stack of spacers subjected to the same procedure, in parallel to the test
objects. For series (B), the moisture content of the impregnated board was
measured by the Karl Fischer titration method, on oil immersed pieces from a
control model running in parallel in the same conditions. The moisture values
refer to the weight of dried insulation.

Test-devices, with model-1 (left) and model-2 (right). The models are
processed under a constant pressure of 5 N/mm2 prior to assembly in
the device. For both model types, the position of the steel plates was
adjusted as to produce initially a load of ideally 1.563 kN (equivalent to
2.5 N/mm2, related to the apparent pressed area of 625 mm2). The test
specimens of series (B) are in 300 ml oil containers (not shown here).
FIGURE 3
Test results
Series (A): Dried, Non-Oiled Specimens In Air
The plot of forces and moisture versus time is given in (Figure 4). The
increase of the clamping force due to the swelling effect of moisture
absorption of the cellulose is prominent. At 3 % moisture content, more than 3
kN was measured for both models. In non-clamped condition, the relative
thickness increase at this moisture level would be more than 2 %.(5) The
different curves of the two model types can be explained by the fact that the
amount of cellulose, in the orientation of the clamping stress, is different: 32
mm for model-1 versus 18 mm for model-2. In addition, wrapped Kraft paper
can be assumed to be mechanically softer than hot-press-dried board.

pressure
(N/mm)

force (kN)
moisture (%)

7
6
5

4
3

4
3
2

2
1

1
0

0
0

10

20

30

model-1 (spacers)

40
time (h)

50

60

70

model-2 (spacers and conductor segments)

moisture

Clamping force and moisture content versus time for the series (A)
non-oiled models. After 24 hours at 23C and 50% relative humidity, the
specimens were dried under vacuum. The N/mm2-scale relates to the
apparent pressure area of 625 mm2.
FIGURE 4

Series (B): Dried and Oil Impregnated Specimens

As expected, the moisture absorption rate of the cellulose immersed in oil was
considerably lower in comparison with the non-oiled insulation of series (A),
(Figure 5). After 450 hours, the moisture had only increased to 0.6 %. Even in
this dry regime, the pressure increase of model-1 is distinct, whereas the Kraft
paper of model-2 seems to settle, and pressure growth is expected to pick up
at a somewhat higher moisture content.

pressure force
(N/mm) (kN)
4.0
2.5

climate chamber

vacuum

vacuum

23C 75 % r.h.

60C

23C

3.5
2
3.0

model-1 (spacers)
model-2 (spacers and
conductor segments)

2.5
1.5

2.0

moisture
0.1 %

moisture
0.3 %

100

200

moisture
0.6 %

300

400

500

moisture
0.2 %

600

time (h)

Clamping force versus time of the series (B) dried, oil impregnated and
oil immersed models. After 450 hours at 23C and 75% relative
humidity, the oil was drained, the oven temperature raised to 60C and
the specimens were dried under vacuum. The steep increase (450 h) is
caused by the thermal expansion of the cellulose insulation (3-fold in
comparison with steel and copper). After 550 hours, the heating was
switched off and the models cooled down to room temperature. The
N/mm2-scale relates to the apparent pressure area of 625 mm2.
FIGURE 5

VI.

EFFECT OF TEMPERATURE

At the moment of the temperature increase to 60C (450 hours), a very

interesting observation was made. The slowly decreasing pressure, as a
consequence of the progressive drying, was strongly masked. The explanation for
this effect can be identified in thermal expansion. It has to be taken into consideration
that any temperature rise causes dimensional expansion of the components of the
test-devices, due to the different material expansion coefficients. The longitudinal
thermal expansion of pressboard ( pb 4510-6/K,
8

perpendicular to the fibre orientation) is a multiple of that of the metal parts involved
(steel st = 1010-6/K, copper: cu = 1710-6/K). This explains why the pressure
increases tremendously with temperature. To prove the theory, the spacers of model1 were replaced by a 30 mm high stack of silicate glass (thermal expansion
negligible). This time, the clamping force decreased from 1.58 to 1.04 kN at a
temperature increase from 23 to 60C, due to thermal expansion of the clamping
structure elements.
The inverse process can be identified at the moment of cooling (after 550
hours). The remaining forces are somewhat lower than prior to the test run.
The pressure increase due to temperature rise is highly dependent on the
materials and the geometry of the clamping structure. In power transformers, the ratio
of clamped cellulose to metal may be different from the present test-devices while the
primary materials: steel, copper and cellulose remain. In order to determine the effect
of temperature on the pressure of power transformer windings one would have to
understand the relationship between the pressure increase due to the expansion of
the materials which compose the clamp, steel clamping structures, and the materials
under the clamping force, copper and cellulose.
By extrapolating the pressure increase recorded during these laboratory test
on models with only cellulose and models with cellulose and copper we can infer what
the effect of temperature change would have on the clamping pressure of transformer
windings. In this example we assumed a 40 mVA two winding transformer rated at
550 kV BIL. The core window height was assumed at 1.5 meters. The ratio of copper
to cellulose in the winding and end insulation was calculated to be 60% copper and
40% cellulose.

WEIDMANN

SK21
60
0
WP
1
Schematic of 550 kV BIL core and coil layout. For the material making
up the winding and end insulation, 60% is copper and 40% is
insulation.
FIGURE 6
Due to the variation in the rate of thermal expansion of the respective
materials there is an effect on the clamping pressure over the operating range of a
transformer.
The thickness of any material will change according to the following formula:
Lt2 = Lt1 (1+ (t2-t1))
10

where:

Lt1 = Thickness of material at temperature t1

Lt2 = Thickness of material at temperature t2
= Coefficient of Thermal Expansion for that material

We know that for materials which are purely elastic according to Hookes law
there is a relationship between the compression in materials and the pressure. This
can be defined by the Modulus of Elasticity.
Thus the formula to calculate the change in pressure due to a change in
thickness of a material is given as follows:
E = (P x L)/ L
where: E = Modulus of Elasticity - N/mm2
P = Pressure - N/mm2
L = Thickness of the material - mm
L = Change in thickness of material - mm
Compression tests on the transformerboard spacer material resulted in a
Modulus of Elasticity value of 500 N/mm2. This value was measured on spacer
material which had been pre-loaded so that it was in the elastic range.
Then for a given transformer, if we know the temperature at which the windings
were pre-loaded and the relative amounts of copper and cellulose in the core
window, we can calculate the change in pressure due to the change in temperature.
Thermal expansion and contraction has an important effect on the factory preloading of the coil assembly. As we have stated earlier, in order to maintain short
circuit withstand capability the windings should be pre-loaded to a pressure at least
equal to the maximum short circuit force. However, it is important at what temperature
this pre-load pressure is applied. In this example if the transformer windings were
loaded to a pre-load pressure of 5 N/mm2 at 90 C (as would be the case
immediately following processing and oil impregnation of the unit) the pressure would
fall to 3.5 N/mm2 at ambient 23 C temperature. If the unit were to be shipped to a
colder climate this pre-load pressure would be reduced even further. (See Figure 7)
These values were calculated assuming that the Transformerboard spacers were
processed following the recommendations given earlier and their behavior is purely
elastic with a Modulus of Elasticity of 500 N/mm2 under compression.

11

Clamping Pressure vs Temperature

6
5.5
5
Pressure (N/mm^2)

4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
-20

20

40

60

80

100

Temperature (Deg. C)

Variation in static clamping pressure of a 550 kV BIL 20 MVA Transformer.

This unit would be clamped at a pressure of 5 N/mm2 at a temperature of
90 C. ( MOE of Transformerboard Spacers = 500 N/mm2)
FIGURE 7

VII.

EFFECTS OF AGING

Over the life of a transformer the cellulose insulation will be subjected to

thermal cycles which will cause aging of the cellulose material. As cellulose ages its
polymer structure degrades into smaller chains which result in the loss of mechanical
strength. One measure of the degradation is the average degree of polymerization.
There has been a substantial amount of work done to measure the effect of thermal
stress on the aging of cellulose insulation. Much of this work has been done in
regards to the effect of aging on the cellulose polymer itself as measured by the
degree of polymerization.(6) (7) The effect of aging on the mechanical properties of
cellulose has also been well documented particularly in regards to the tensile strength
and burst strength properties of the cellulose.( 8) We have found in recent tests that the
aging of cellulose material has an even more pronounced effect on the thickness of
material under pressure. The effect of aging is more pronounced due to the dual
effect of material decomposition due to degradation of the cellulose polymer (depolymerization) but also to the settling of the material due to pressure.
12

In a recent aging test, 100 mm high stacks of 3 mm thick pieces (25 x 25 mm)
of high density pre-compressed pressboard were placed under a constant load
(spring load) of 10 N/mm2, at temperatures of 135oC and 150oC for 36 weeks = 250
days of aging. The aging vessels were operated in an open expansion system in
which the oil in the expansion vessel is continuously in contact with dry air via a silica
gel dryer. The design of the aging vessel and the materials within the vessel was
done to simulate the conditions seen in a transformer. Relative surface areas of the
vessel surface, core steel, copper, cellulose insulation and oil were calculated to
simulate those in a 200 MVA transformer. The pre-compressed transformerboard
spacers were dried and oil impregnated under a pre-load pressure of 10 N/mm2 prior
to being loaded into the aging vessels.
The temperature of the aging vessels was cycled daily to simulate the load
cycle that a typical transformer might see. The temperature cycle for a 24 hour period
composed of the constant temperature ( 135C or 150C) for 19 hours , followed by a
2 hour cooling phase and then by a 3 hour heating phase. The loss of effective aging
time based on the Montsinger half value of 8 K was about 4 hours per temperature
cycle per day. (9)
The pressboard shows increasing shrinkage caused by loss of substance
(CO2, CO, H2O) during the aging of the cellulose and decrease in mechanical
strength (due to decomposition of cellulose molecules). The effect of aging on the
thickness of pre-compressed cellulose board as a result of this experiment is shown
in (Figure 8). These values verify earlier aging tests performed at Weidmann.(9)

13

Aging of Pressboard Under Compression

Spacer Stack Height (mm)

102
100
98
96

135 Deg. C

94

150 Deg. C

92
90
88
0

50

100

150

200

250

300

Aging Time (Days)

Effect of aging on the thickness of a stack of Transformerboard.

Samples were aged in an open expansion system via silica gel dryer
with daily temperature cycling.
FIGURE 8

The following table (Table I) gives the % thickness change due to loading, drying
and aging for 250 days. Also included is the expansion after pressure is reduced
from 10 to 1 N/mm2.

TABLE I
Static compression

Thickness Change

Loading shrinkage 1.. 10 N/mm2

-3,0

Drying shrinkage

-7,5

Aging shrinkage

135 C

-4,8

after 250 days

150oC

-11,0

Unloading expansion 135oC

+1,1

150oC

+0,7

10 1 MPa (after
250 days of aging)

14

In this aging study the traditional property of Degree of Polymerization was

measured at the end of the aging period of 250 days.

DP

Initial values

135oC

150oC

1190

164

152

Of interest in this aging study is the large difference in the shrinkage between
the samples aged at 135 C and the samples aged at 150 C, while the traditional
values of DP does not show this. The influence of aging on the thickness of the
spacers under compression then must be a more complex phenomena. The data
demonstrate that while degree of polymerization appears to have leveled off at or
below a value of 200 that would indicate the end of life, the thickness of the spacer
material continues to decline.
VIII.

CONCLUSION
Manufacturing

In order to assure that a transformer is capable of withstanding the forces

which are generated during a short circuit event, the windings must be rigidly
clamped. A common practice in the design and manufacture of power transformers is
to clamp the windings to a pressure which is at least as great as the maximum short
circuit force. The transformerboard spacer material itself should be of pure high
quality kraft cellulose and have the surfaces milled to avoid excessive compression
set. The information presented in this paper gives further insight into critical process
parameters which must be regarded in order to assure proper pre-loading of the
transformer windings. It is critical that the transformer insulation structure is thoroughly
dried and oil impregnated prior to the final pre-loading pressure being applied.
Should a transformer winding be clamped after the insulation has reabsorbed
moisture, the windings will be at a reduced pre-load pressure after processing or
during operation.
The coefficient of thermal expansion for cellulose insulation material is
approximately three times the value for copper and steel. The effect of temperature
on the pre-load pressure has been demonstrated. It is best to pre-load the
transformer at 30C to 50C in order to assure that the windings have a proper preload force over the operating range of the transformer. This is particularly important in
climates where the ambient temperature can be quite low in order to avoid loss of
correct pre-load pressure.

15

Service
During the service life of a transformer, many interesting phenomena take
place in the insulation structure which will affect the pre-load pressure of the windings.
The moisture of the cellulose insulation will increase over time due to a number
of factors. The aging process of the cellulose has water as a primary byproduct. The
increase in moisture in the cellulose insulation will have a tendency to expand the
insulation resulting in an increase in the pre-load force.
The transformer will go through daily and seasonal load cycles which results in
temperature changes as well. The effect of temperature on the pre-load pressure has
been demonstrated in this paper. The temperature effect on pressure during
operation could be offset by the moisture dynamics of the cellulose/oil system.(10) As
the transformer cools the moisture will have a tendency to migrate from the oil into the
cellulose insulation. As we have shown an increase in moisture content will cause the
cellulose insulation to expand thus offsetting the pressure loss due to the lower
temperature. The time constant of the moisture migration can have a significant
impact on this phenomena. While this has not been studied in this investigation it is
intuitive that the thermal expansion effect will have a lower time constant than the
moisture migration.
Increased moisture in the insulation system has a significant effect on
increasing the winding pre-load pressure, regardless of load. Increased loading and
operating temperatures result in aging and exaggerated loosening of the windings.
On occasion, for units with high loading and high moisture, the moisture tends to hide
the loose winding problem, until such time as the unit is reprocessed. The effect of
aging on the thickness of cellulose transformerboard has been shown. The reduction
in thickness due to aging is twofold. Aging degrades the cellulose polymer and also
results in increased set on the transformerboard. Any operation at temperatures
greater than 100C will result in a loosening of the transformer windings due to aging.
It is a well known fact that transformers removed from service at or near their end of
life have been found to have little or no pressure remaining on the windings.(11) This
observation is not always possible, because when the unit is allowed to reabsorb
moisture after being removed from service, the windings can appear tight and lead to
the false conclusions that they were tight during operation.

Field Processing of Transformers

16

As the insulation structure of a transformer ages it will generate H2O ,CO and
CO2 as primary byproducts. Moisture will act as a catalyst to enhance the aging
process even further. Due to the trend toward a higher level of acceptable transformer
loading (12) and in order to prolong transformer life, many users are more frequently
monitoring and processing the oil of units in the field. This removes the excess
moisture and other byproducts of aging. The removal of moisture will improve the
overall dielectric system and extend insulation life. However, it will have a detrimental
effect on the remaining pre-load pressure of the transformer windings.
To assure that transformer windings maintain sufficient pre-load pressure,
particularly in larger units, the windings may require inspection and tightening in the
field. The unit should be processed prior to inspection to assure that the pressure
remaining on the windings represents that of operating conditions. If tightening is
found to be necessary, it is important that the pressure is applied uniformly over the
windings, and in humid conditions, as quickly as possible after reprocessing to avoid
uptake of moisture.

REFERENCES
17

1.

K. Karsai, D. Keranyi, L. Kiss Large Power Transformers, Studies in

Electrical and Electronics Engineering. Elservier 1987 pp. 570-571.

2.

H.P. Moser, Transformerboard , H. Weidmann AG, 1979, pp. 71-77.

3.

Ch. Krause, W. Goetz , The Change of the Clamping Pressure in

Transformer Windings Due to Variation of the Moisture Content - Tests with
Pressboard Spacer Stacks , CIGRE SC 12 Transformers / Workshop on
Short Circuit Performance of Transformers - 1999 Budapest Colloquium.

4.

G. Heldmaier, "Spacerboard - For Short Circuit Resistant Transformers",

Weidmann Bulletin, June 1996.

5.

Ch. Krause, H.P. Gasser et al., "Effects of Moisture in Transformerboard

Insulation and the Mechanism of Oil Impregnation of Voids", Transform 98
Symposium Munich, 1998.

6.

W.G. Lawson, M.A. Simmons, P.S. Gale, Thermal Ageing of Cellulose Paper
Insulation, IEEE Transactions on Electrical Insulation, Vol EI-12, No. 1,
February, 1977.

7.

H.P. Moser, V. Dahinden, "Transformerboard II", Section S, H. Weidmann AG,

1987, pp. 144 and 149.

8.

T.V. Oommen, L.N. Arnold, Cellulose Insulation Materials Evaluated by

Degree of Polymerization Measurements IEEE Transaction Paper 1981.

9.

H.P. Moser, V. Dahinden, "Transformerboard II", Section S, H. Weidmann AG,

1987, p. 140.

10.

J. Fabre, A. Pichon, "Deteriorating Processes and Products of Paper in Oil.

Application to Transformers", CIGRE Session, Paper No. 137, 1960.

11.

W. J. McNutt, G.H. Kaufmann, Evaluation of Life Test Model for Power

Transformers, IEEE Transactions on Power Apparatus and Systems, Vol.
PAS-102, No. 5, May 1983.

12.

D. J. Woodcock and J. C. Wright, PE., Power Transformer Design

Enhancements Made to Increase Operational Life, Sixty-Sixth Annual
International Conference of Doble Clients, April 12-16, 1999.

18