2011 Electrical Insulation Conference, Annapolis, Maryland, 5 to 8 June 2011

High Voltage Power Transformer Insulation Design

W.Ziomek, K.Vijayan, D.Boyd, K.Kuby M.Franchek

Engineering Department Weidmann Electrical Technology Inc.
CG Power Systems Canada Inc St. Johnsbury, VT, USA
Winnipeg, MB, Canada

Abstract— The large power transformer life span depends uniform theory explaining all processes leading to breakdown
heavily on insulation – its condition, materials, composition, in oil. Theoretical studies established the following main
geometry, etc. The high voltage transformer insulation requires a mechanisms of breakdown in liquids: (i) electrical, with
very focused analysis in the design stage. The large power development of electron avalanche, leaders, and streamers, (ii)
transformer is subjected to different overvoltages during factory ionic, utilizing the ion conduction in contaminated liquid, (iii)
testing and in operation, therefore all resulting electric stresses suspended particles, which polarize in the field and
have to be modeled in the electric field program and compared to concentrate, resulting in breakdown, (iv) gaseous, where
the industry accepted withstand curves. The overvoltages include presence of gas bubbles decreases the local dielectric strength,
transient voltages from impulse testing, both lightning and
(v) electro-convection, involving dynamics of space charge in
switching, as well as power frequency voltages generated during
liquid and charge deposition on the cellulose insulation, see e.g.
the induced and applied voltage tests. The distributions of electric
stresses under these different electric excitation conditions are [1-10]. For practical purposes of estimating the withstand of
different. The transient voltage programs with MLCR insulating mineral oil, somewhat simplified approach is
components (mutual- and self-inductance, capacitance, suggested [2, 3], where the dielectric strength was expressed as
resistance) are used to establish exact voltage distribution inside a function of (i) interelectrode gap, (ii) surface of electrodes,
the transformer. The electric field analysis for complex units is and (iii) the volume of oil contained between electrodes, where
performed for all energized components, including turn-to-turn, the latter is a result of two former phenomena. The volume
section-to-section, winding-to-winding, winding to core, phase–to- effect [2] allows for comparison of the breakdown strength in
phase, between leads, from winding and leads to constructional different electrode systems. The method utilizes the
parts, bushing to tank, etc. The electric stresses are checked equigradient areas in the interelectrode space, i.e. surfaces of
under two main conditions: (i) strike in oil, (ii) creeping discharge same dielectric stresses (kV/mm). It appears that the equivalent
along solid insulation. The paper will discuss the analysis for volume of stressed oil for spherical electrodes contained within
selected transformers and explain the design process needed to ~90% equigradient surface corresponds to withstand of plane
maintain the dielectric stresses under critical values. electrodes of same volume. Numerous experiments [2,4,11,12]
led to conclusion that the dielectric strength of oil gap, Ebd, was
Keywords: high voltage insulation, power transformer, partial a power function of the equivalent volume, Veq.
discharge, insulation design
In general, the dielectric strength of an oil cellulose
I. INTRODUCTION insulation system depends on the duration of voltage
application, polarity of voltage, field enhancement factor, area
The modern electric power industry employs two basic and shape of electrodes, kind and degree of contamination of
transformer constructions: (i) core-type, which is predominant, the oil, its temperature and pressure. The transformer insulation
and (ii) shell-type. This paper will focus on the insulation design should be prepared with careful consideration for all
system for core-type transformers with cellulose-oil insulation. these aspects.
Insulation for the core-type, oil-immersed power transformer
consists of (i) the main insulation, typically utilizing mineral oil
(also acting as the cooling medium) and pressboard barriers in II. BREAKDOWN IN OIL CRITERIA FOR INSULATION DESIGN
winding to winding, winding(s) to ground, lead to lead, lead to The subject of breakdown in oil under different voltages
ground spaces, and (ii) the winding insulation: section to has been intensively researched for many decades and resulted
section, turn-to-turn, lead to winding. Materials commonly in establishing the breakdown curves [3,5,6,8,9,10,13]. In
used in insulation system are: (i) insulating fluid: mineral oil, concise summary one may collect these results in groups as
synthetic, or vegetable oil (natural esters), (ii) conductor follows:
insulation: paper (kraft, Nomex®, enamel), (iii) ‘solid’
insulation, i.e. barriers, blocks, spacers: pressboard • Breakdown curves for oil gaps (the strike in oil
(Weidmann’s nomenclature: transformerboard TIII, TIV), curves), developed in years 1950-80 [1, 2, 5, 13], and
transformerwood (densified wood), wood (typically it is maple known as Weidmann curves (Fig.1). These curves are
or beech wood). based on the 1% probability of partial discharge
inception in the uniform oil gap under ac power
The development of a breakdown in liquid is complex and frequency voltage, corresponding to a 1 minute applied
still under ongoing investigation. At present, there is no voltage test.

978-1-4577-0279-2/11/$26.00
978-1-4577-0276-1/11/$26.00 ©2011 IEEE 211
Authorized licensed use limited to: Centro Universitário Fei. Downloaded on April 11,2024 at 13:48:12 UTC from IEEE Xplore. Restrictions apply.
 • Creep breakdown curves were initially based on the oil The impact of moisture on long term operation of
gap breakdown curves [5, 6, 10], later improved transformer insulation is one of the most critical factors.
through laboratory research (Fig.2). Knowing that the electric power industry accepts the water
content in oil as high as 30 ppm, the paper-oil moisture
• The volume effect curves are used when the unit under equilibrium curves (Fig.3) give moisture in paper reaching 6-
analysis is extremely large, or the voltage is distributed 7%. Such high content of water will decrease withstand of
over multiple parallel branches resulting in stressing of paper insulation by 50-70% (see Fig.4). Therefore, in most
a larger volume [2, 11, 12]; cases, the withstand level for paper should be established based
on the operating conditions rather than on test conditions.
Electrical Strength of Transformer Oil
100
IV. INSULATION SYSTEM DESIGN
Partial Discharge - Inception
Field Strength (kV/mm)

Design of main insulation system, especially winding-to-

winding, winding(s) to core, lead exit, etc., is realized through
10 the analysis of electric field distribution. The stresses obtained
from transient voltage analysis are converted to equivalent ac
voltages, these voltages are applied to the geometry of the
1 insulation system under analysis and resulting electric stresses
0.1 1 10 100 1000 are compared to industry accepted withstand curves [5, 10, 12].
Oil Duct Width (mm) The fundamental relationship between different transient
voltages lightning impulse full wave (LI FW), switching
Degassed Oil, Non-Insulated Electrode
impulse (SI) and ac test voltage are shown in Table I. These
Degassed Oil, Insulated Electrode relationships have been developed over years of high voltage
laboratory testing.

Figure 1. Weidmann curves for strike in oil Design of the winding insulation is based on experimentally

• Evaluation of characteristics of non-standard impulse

waveshape, e.g. impulses of short duration, or long
duration [14], establishing the equivalent lightning
impulse (LI) or ac test level for these impulses.

Figure 3. Moisture equilibrium curves for paper-oil system

60

50
x = 1%
Voltage U(kV)

40
x = 4%
30 x = 6%
Figure 2. Creep withstand curve (1% probability of partial discharge x = 8%
inception) developed by EHV Weidmann 20
x = 10%
10

0
30 40 50 60 70 80 90 100
III. IMPACT OF SELECTED FIELD CONDITIONS ON Temperature (°C )
WITHSTAND LEVEL OF INSULATION
Figure 4. Reduction of dielectric strength of paper insulation as a
The dielectric strength of oil insulated system depends function of moisture content for 1 mm paper
mainly on: (i) duration of voltage application, (ii) kind of
developed withstand curves, typically established by individual
voltage (waveshape), (iii) polarity of voltage, (iv) field
manufacturers. The ‘traditional’ approach uses different
enhancement factor, (v) geometry: distance, area and shape of
breakdown curves (e.g., 1% probability) for basic winding
electrodes, (vi) kind and degree of contamination of the oil
types, e.g. continuous disc winding, interleaved winding; the
(e.g. water), its temperature, velocity and pressure [12, 15, 16].
oscillations generated during testing for given winding type

212
Authorized licensed use limited to: Centro Universitário Fei. Downloaded on April 11,2024 at 13:48:12 UTC from IEEE Xplore. Restrictions apply.
will correspond to those used in preparation of breakdown The local dielectric stress at the surface of an insulated
curves (similar duration, oscillations, stress distribution). The electrode, such as a winding section, cannot exceed 11-12
more ‘universal’ approach consists of two stages: kV/mm. The surface stress level for an un-insulated electrode
(i) obtaining different breakdown curves (1% probability) for or a ground plane is significantly lower and may need to be
different voltage-time characteristics, e.g. full wave, chopped below 2kV/mm. This local stress is affected by not only the
wave, unipolar oscillation, bi-polar oscillation, applied voltage but also by the geometry factor, which is
(ii) model transient voltages in transformer and compare the related to the electrode shape. This can increase the local
node voltages to appropriate withstand curves. As example, in surface stress significantly. Figure 7 below depicts a typical
Fig.5 the 1% probability breakdown stresses are presented for FEA field plot showing the equipotential lines. The maximum
varying duration of applied LI. This method requires quite local stress on this coil geometry occurs on the 4th section from
the coil static plate. This area is shown with a red X on Figure
100
7. The maximum local stress many times does not occur at the
1% probability breakdown field

coil ends, but in an area where there is a geometric feature that

enhances the stress level. In this example the high stress was
stress at LI [kV/mm]

LI 0.5/1.5 μs
caused by a combination of high impulse voltages and coil
LI 1.0/2.0 μs geometry. Since the dielectric stress exceeds 11kV/mm an
LI 1.2/50 μs additional solid insulation component is required to reduce the
stress in the oil.

10
0 2 4 6 8 10 12
Distance between sections [mm];
at 3 mm paper covering

Figure 5. 1% probability breakdown stress between two sections with

3mm paper covering on winding conductor for varying duration of
voltage waveshape (LI front/half-tail: 1.2/50, 1.0/2.0, 0.5/1.5 s)

Table 1. Design Insulation Factors (DIL) in reference to the

ac power frequency 1 minute test

Condition DIL factor

Lightning impulse (FW) 2.3-2.7
Switching impulse (SS) 1.8-1.9
AC one-minute test level 1.0
AC one-hour test 0.8
AC operating voltage 0.55-0.59 Figure 6. Typical FEA solution – coil ends

extensive laboratory work on preparation of breakdown curves,

but leads to safer analysis of stresses within the winding. In the FEA field plot many of the insulating oil gaps are
evaluated for the stress levels across the oil gap. The actual
V. INSULATION SYSTEM DESIGN – EXAMPLES stress in the gap is then compared to the oil strength curve
shown in Figure 1. If the actual stress across the oil gap
The traditional insulation system design for a high voltage exceeds the allowable design stress, an additional solid
core form transformer begins by modeling the transformer cellulose barrier would be required in the insulation system to
insulation system in either two or three dimensions (see figures increase the oil gap strength. The insulating barriers are used
6 and 12). The decision to use a two- or three-dimensional to control the dielectric stress in all of the transformer windings
model is usually based on the complexity of the geometry and by dividing the oil into short distances (smaller volumes).
the overall detail required to obtain an accurate solution to the
dielectric field stresses. These models are sometime completed
in the finite element program but in many cases are prepared in
Figure 8 shows an example for the method to determine
a CAD system then transferred to the finite element analysis
(FEA) software. the dielectric stress across an oil gap in the insulation system.
The actual stress is calculated across the oil gap from point A
to point B.
There are three critical criteria that must be analyzed in
completing and optimizing the insulation system design, they
include: (i) maximum local stresses at the surface of an
insulated or un-insulated electrode, (ii) the stresses across the
oil gaps and (iii) the creep stresses along the solid insulation.

213
Authorized licensed use limited to: Centro Universitário Fei. Downloaded on April 11,2024 at 13:48:12 UTC from IEEE Xplore. Restrictions apply.
 Figure 7. Local maximum stress < 11kV/mm

Figure 9. Comparison of the actual stress to design curve stress in oil gap A
to B.

Figure 8. Determination of the dielectric stress across an oil gap at the end of More complex systems, such as a bushing insulation in
the HV coil vicinity of windings and grounded components, require a 3D
field analysis (Fig.12) with subsequent stress analysis on
The stress calculated across the A to B oil gap (between a critical planes identified in a 3D model (Fig.13). In larger
stress ring and a cellulose angle ring barrier) is compared to the transformers once all of the areas of the insulation system are
design criteria as shown in Figure 9. From this design curve a evaluated for local, oil withstand and creep stresses, a review of
percent margin is calculated to determine the overall safety the stressed oil volumes would also be completed. The process
factor in the insulation system design. In the field plots the of optimizing the insulation system is an iterative process that
color palette denotes relative stress levels in the insulation typically requires 4 to 8 iterations. The iterations involve
system; the blue areas being the lowest stress and the red being adding, omitting and rearranging the solid insulation barriers to
the highest. It should be noted that the solid insulation barriers maintain an acceptable percent design margin throughout the
are lower in stress due to the higher permittivity of the insulation system. The final optimized insulation system design
transformerboard than that of the oil or paper. will be a balanced design with minimum required dielectric
The final criterion that the insulation system must meet is clearances and solid insulation components.
that any tangential creep stresses along the interface surface
between the solid and oil insulation may not exceed the design
limit. The creep stress design curve is shown in Figure 2. This
requires analysis of the stress along any long insulating
component which has significant tangential stress. This type of
analysis is shown in Figure 10. The resulting stress profile
along the solid insulation surface is typically not monotonically
decreasing in value, as is typical for oil gap stresses. A typical
stress profile is shown in figure 11. This creep stress profile is
converted to a monotonically decreasing function by
processing the stress profile using a cumulative integration
algorithm. The resulting stress curve can then be compared to
the creep strength curve shown in Figure 2.

Figure 10. Creep stress analysis

214
Authorized licensed use limited to: Centro Universitário Fei. Downloaded on April 11,2024 at 13:48:12 UTC from IEEE Xplore. Restrictions apply.
 VI. CONCLUSIONS
Insulation design for power transformers need to consider
many parameters including type of voltage, test connection,
electrode geometry, transient voltage at the node, stressed area
and withstand characteristics of cellulose-oil system. Details of
insulation design procedure and optimization method are
discussed. Typical insulation design and FEA analysis for
various stress conditions at critical locations of a transformer
are included. Example for 3D FEA analysis for a complex
geometry is also included.

Figure 13. 3D FEA field plot lead exit bushing and HV winding

REFERENCES
[1] Blume L.F. et al „Transformer Engineering”, J.Wiley and Sons,New
York, 1959
[2] Wilson, W.R. „A Fundamental Factor Controlling the Unit Dielectric
Strength of Oil”, AIEE Transactions, Feb.1953, pp.68-74
[3] Kappeler, H. „Recent Forms of Execution of 380kV Transformer
Bushings, CIGRE, paper No.126, 1958
[4] Hasterman Z., Mosinski F., Maliszewski A „Electrical strength of power
transformers” (in Polish), WNT, Warsaw, 1983
[5] Moser et al „Transformerboard”, 1979
[6] Krause, Ch. „Dielectric Strength of Transformer Oil”, in Proc. of
WICOR Insulation Conference, Rapperswil, Switzerland, Sept.1996
[7] Kuffel, E., Zaengel, W.S., Kuffel J., High Voltage Engineering,
Butterworth-Heinemann, Newnes, Oxford-Boston-Melbourne, 2000
Figure 11. Creep stress cummulative analysis [8] Nelson, J.K. „An Assessment of the Physical Basis for the Application
of Design Criteria for Dielectric Structures”, IEEE Transactions on
Electrical Insulation , Vol.24, No.5, Oct.1989
[9] Lokhanin A.K., Morozova T.I., et al „Problems of Coordination of
Dielectric Strength of Extra High-Voltage Power Transformer Major
Insulation, CIGRE, 12-06, 1970
[10] Tschudi, D.J., „AC Insulation Design”, in Proc. of WICOR Insulation
Conference, Rapperswil, Switzerland, Sept.1996
[11] Giao Trinh N., Saker A., „Interpretation of the Direct Breakdown
Mechanism in Long Gaps in Transformer Oil”, Proc. of 10th
Int.Symposium on High Voltage Engineering, Montreal, Canada,
Aug.25-29, 1997
[12] Siodla K., Ziomek W., Kuffel E. “The Volume and Area Effect in
Transformer Oil”, Proc. of 2002 IEEE Symposium on Electrical
Insulation, Boston, USA, May 2002.
[13] Knorr, W., Breitfelder, D., „Breakdown in Transformer Oil with AC and
Impulse Voltage Stress“, IEEE Symp. on El. Insulation,Montreal, 1984
[14] Okabe, S., Takami, J., “Evaluation of Breakdown Characteristics of Oil-
immersed Transformers under Non-standard Lightning Impulse
Waveforms - Method for Converting Non-standard Lightning Impulse
Waveforms into Standard Lightning Impulse Waveforms”, IEEE
Transactions on Dielectrics and Electrical Insulation Vol. 15, No. 5;
Oct.2008
[15] Del Vecchio R.M., Geometric Effects in the Electrical Breakdown of
Transformer Oil, IEEE Transactions on Power Delivery, Vol.19, No.2,
April 2004
[16] Balma P.M., Degeneff, R.C.,Moore H.R., Wagenaar, L.B., The Effects
of Long Term Operation and System Conditions on the Dielectric
Figure 12. 3D model and FEA plot lead exit bushing and HV winding
Capability and Insulation Coordination of Large Power Transformers,
IEEE Transactions on Power Delivery, Vol.14, No.3, July 1999

215
Authorized licensed use limited to: Centro Universitário Fei. Downloaded on April 11,2024 at 13:48:12 UTC from IEEE Xplore. Restrictions apply.