XVth International Symposium on High Voltage Engineering

University of Ljubljana, Elektroinštitut Milan Vidmar, Ljubljana, Slovenia, August 27-31, 2007 T8-329.pdf

Evaluation of the dissipation factor under different environmental conditions

M. Muhr, R. Schwarz and C. Sumereder
Institute of High Voltage Engineering and System Management
Graz University of Technology
Inffeldgasse 18, 8010 Graz, Austria
[email protected]

Abstract: For the extension of life time and under the dielectric properties of high voltage insulation of
aspect of reliability diagnostic measurements represent a electric energy equipment are to be assessed. A rise of
fixed part in the choice of the maintenance strategy. The tan δ is a sign for strong worsening of the insulation
diagnostic measurement can be distinguished to condition, mostly caused by internal partial discharges.
electrical, mechanical, chemical and thermal methods. The classical circuit for the measurement of capacity
For electric power generation, transmission and and dissipation factor is a “C / tan δ” Schering Bridge,
distribution systems the dielectric diagnosis is applied. which is an AC-balanced bridge circuit. It is
Often used measurements are: insulation resistance, characterised by the fact that the object reality is
capacitance, dissipation factor and dielectric response stressed closed to high voltage, which can be measured
methods in time and frequency domain (IRC, PDC, (opposite usual alternating voltage bridges). A
RVM, FDS, …). The advantage is the destruction free balancing process of the bridge is required.
application with the restriction that the measurement is Recently new electronic bridges were developed
only off-line possible. where no balancing of the bridge is necessary. This
The dissipation factor represents an imported parameter capacitance and dissipation factor bridges are computer-
for the condition evaluation of high voltage power based measuring systems which work according to the
equipment. The classical method to determine the principle of a vectorial impedance measurement in the
dissipation factor and the tan δ tip -up is the Schering frequency range by analysing the fundamental harmonic
Bridge. New computer-based measuring systems of the currents of the test object and the reference
operate according to the principle of a vectorial capacitance [1]. So these bridges measure the ratio of
impedance measurement in the frequency range by their input currents. Thereby the dissipation factor
analysing the fundamental current harmonic. Earlier results from the measurement from the phase shift of the
investigations have shown a very good correlation current signals of a measuring branch and the reference
between the results of both methods. branch.
In most cases at on-site measurements different The comparison branch consists of a reference high
environmental conditions (primary temperature and air voltage capacitance in series with the input (low voltage
humidity) were given. For a condition evaluation a terminal) of the current sensor of the electronic bridge.
direct comparison of test results is necessary, due to this The measuring branch consists of the test object in
fact the values have to be corrected. In this paper series with the input of a second current sensor. The
dissipation factor measurements under different measuring principle is schematically shown in Fig. 1.
environmental conditions were carried out. In special
the behaviour at different temperatures and/or air
humidities were simulated in a climate chamber under
Referenze Test
high voltage. The results were evaluated and a capacitor object
correction factor was investigated. As test object
generator bars (resin rich / mica insulation system) were
used.
CN CX
tan δN IN IX tan δ X
1 INTRODUCTION
Dielectric diagnostics observe and evaluate the
effect between the electromagnetic field and the
material. Insulation resistance, Capacity and dissipation I X C X 1 + j tan δ N
= ∗
factor as material and device-specific parameters give I N C N 1 + j tan δ X

significant information about changes in the insulation

medium.
The dissipation factor and/or capacity measurement
are important methods for the actual condition Fig. 1: Capacitance and dissipation factor measurement
evaluation of insulating systems especially when the system at high voltage

1
 The amplitude und phase of the currents were
measured, amplified digitised and transmitted to the
computer. The value of the capacitance and dissipation
factor of the reference branch is necessary. With this
data the dissipation factor tan δX and the capacitance C X
of the test object can be calculated by applying a
Discrete Fourier Transformation.
Fig. 3: Preparation of generator bars for tan δ measurements
2 THE DISSIPATION FACTOR AND ITS
DEPENDANCE PARAMETERS 3.1. Temperature
Doing measurements onsite the environmental The permanent operating temperature was taken
conditions like temperature and air humidity were given with an infra red measuring device. The temperature
and can not be influenced. The dissipation factor of started at room temperature and was raised in steps up
electrical insulation systems is strongly dependant to to the maximum rated operating temperature of 90 °C.
these parameters and is varying with time (summer, The dissipation factor measurements were done with a
winter or morning, afternoon) and with load (low load, LEMKE LDV6 between 20% and 120% of rated
full load). For this reason the following measurements voltage at each temperature step. The results were given
should give a view how much the dissipation factor can in Fig. 4.
shift due to temperature and air humidity.
0.08

3 TEST SETUP AND RESULTS 0.07

tan δ = f(T)
Dissipation Factor [%]

0.06

In the first test a generator bar of resin rich type was 0.05 12kV
taken to measure the temperature dependence. In the 0.04 10kV
8kV
second test series the behaviour of tan δ with air 0.03
6kV
humidity variation was investigated. The measurements 0.02
4kV
were done in a climate chamber where temperature and 0.01 2kV
humidity could be defaulted, see Fig.2. 0
20 30 40 50 60 70 80 90 100
Test Temperature [°C]

Fig. 4: Tan δ in dependence of temperature

In Fig. 4 it can be clearly seen that there is a strong
temperature dependence which is not linear. Doing a
mathematic approximation of the measured curves
following polynomial functions are fitting very exactly
the measured results:

2kV: tan δ (T) = 0,0148 - 0,0001.T + 8.10-6 . T² (1)

12kV: tan δ (T) = 0,0307 - 0,0004.T + 9.10-6 . T² (2)

The dependence of test voltage in each temperature

step was analysed and the result is shown in Fig. 5.

0.08

0.07 tan δ = f (U)

90°C
0.06
82°C
Dissipation Factor [%]

0.05 71°C
0.04 62°C
53°C
Fig. 2: Climate chamber 0.03
45°C
0.02 35°C
23°C
The preparation of the test objects can be seen in 0.01

Fig. 3. All dissipation factor measurements were done 0

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
with air gap and guard rings to prevent of leakage Test Voltage [V]
currents.
Fig. 5: T an δ in dependence o f test voltage

2
 In Fig. 5 can be seen that the dissipation factor for 3.2. Air humidity
each temperature corresponds to a linear function and
The second parameter of high interest is the air
the voltage dependence is low. For a mathematical
humidity. Air humidity can vary in dependence form air
description a linear equation can be used:
temperature very wide. Normally relative air humidity
between 30 and 80 % is given, in extreme climates
23°C: tan δ (U) = 0,0115 + 1.10 -6 . u (3) (desert, tropical) lower or higher values can be existent.
90°C: tan δ (U) = 0,0623 + 1.10 -6 . u (4) For this test series different types of generator bars
were tested in a climate chamber, where the temperature
In literature the temperature relation of the was constant and the air humidity could be chosen
dissipation factor is often given as an exponential between 20 % and 98 %. The measurements were only
function. In [2] a comprehensive literature review was done when the state of equilibrium was reached, the
done and following mathematical formulation was climate conditions were measured with a TESTO 400
found: system.
The test objects were of resin rich (RR) and vacuum
tan δ (T R) = tan δ (T M) . (e -k (TM -TR)) (5)
pressure impregnation (VPI) type, also different glass
with TM … temperature during measurement fibre tapes were used. Four test objects could be
TR … reference temperature with known tan δ observed parallel, so for these four bars identically
k … system factor conditions were given. As example of evaluation some
representative diagrams should be shown. In Fig. 7 the
This formula requires one known value of tan δ and dissipation factor for a VPI system with a standard glass
the k-factor. With a k-factor of 0,017 [2] equation 6 fibre tape was tested. The diagram shows the tan δ for
results. This equation is plotted in a diagram, shown in different test voltages, the results were within very close
Fig. 6. divergence.

0.08 0.14

tan δ = f (T)
0.07
0.12
Dissipation Factor [%]

0.06
0.1
Dissipation Factor [%]

0.05
0.08 TD@100V
0.04 TD@500V
0.06 TD@1kV
0.03
0.04
0.02

0.01 0.02

0 0
20 30 40 50 60 70 80 90 10 20 30 40 50 60 70 80 90
re la tiv e air humidity [ %]
Temperature [°C]

Fig. 6: C alculated tan δ in dependence of temperature Fig. 7: Tan δ in dependence of air humidity for a VPI System

Comparing the calculated tan δ from equation 6 with At lower air humidity the tan δ is constant at 2%,
the measured curve of Fig. 4 the best correspondence from 53% relative humidity a sharp bend can be
can be found for a test voltage of 6kV. The polynomial observed. The curve increases linear up to a tan δ of
function (equations 1, 2) was transformed to an 12% at 85% r.h. Same observations were done at the
exponential function as equation 5, the result can be other VPI bars.
found in equation 7. The divergence of both equations The resin rich bars showed a similar behaviour,
was within very close borders and test results meet the starting with a constant value of 2% the tan δ raises
specification from [2] exactly. almost linear from a humidity of 59% r.h. up to 16%.
RR FI
tan δ (measured) = 0,0139 . e0,0172T (6) 0.2
tan δ (calculated) = 0,0151 . e0,017T (7) 0.18
0.16
Dissipation Factor [%]

0.14
For condition evaluation the dissipation factor 0.12 TD@200V

should always be taken at a reference temperature, 0.1 TD@500V

0.08 TD@1kV
mostly at room temperature. With this equation the 0.06
dissipation factor at different temperatures can be 0.04
0.02
calculated. This means that it is not necessary to wait 0
until the machine cooled down to the reference 10 20 30 40 50 60 70 80 90 100

temperature, a long period of time can be saved. relative air humidity [%]

Fig. 8: Tan δ in dependence of air humidity for a RR System

3
 To explain the bend in the curve further 10
investigations were done. The sudden raise of the 9

dissipation factor could be caused by a change of the 8

insulation resistance. For this reason the surface and the

Surface Resistance [TOHM]

7

volume resistance at same environmental conditions 6

were m easured with a Megohmmeter. Surface and 5

4
volume resistance can not be measured separately, the
3
cross section and equivalent circuit can be found in
2
Fig. 9. 1

0
10 20 30 40 50 60 70 80 90 100
relative air humdity [%]

Fig. 11: Surface resistance of RR System

The surface resistance is linear decreasing until a
humidity of approximately 50% is reached. Resuming
the results of the insulation resistance measurement it
can be pointed out that between 50 and 70 % a change
in the behaviour of the insulation medium can be
Fig. 9: Cross section of a generator bar and equivalent circuit
observed. The characteristic of the dissipation factor
for insulation resistance measurement also changes in the same range.

The measurements were done in two steps: first the

4 CONCLUSIONS
insulation current on the semiconductive coating (A)
was measured for the approximation of the volume
Electrical insulation systems show a strong
resistance, guard rings should prevent surface currents
dependence to temperature and air humidity. For the
(B grounded). And second the insulation currents
behaviour of each parameter several measurements were
between copper and guard rings (B) were measured for
done and mathematical models were developed.
the approximation of surface resistance. The high
The temperature dependence was reduced to an
voltage (5kV DC) was applied to the conductor.
exponential function. With the described function it is
10
possible to calculate the dissipation factor at any
9 temperature if the system parameters were known.
8
The dependence of the dissipation factor from air
Volume Resistance [TOHM]

7
6
humidity is more complicated and it has been shown
5 that the surface currents over the voltage grading tape
4 has important influence. From a relatively air humidity
3
2
of 50% the surface resistance drops rapidly and
1 influences the measurement results. As recommendation
0
10 20 30 40 50 60 70 80 90 100
for dissipation factor measurements the influence of the
relative air humidity [%] air humidity should be observed carefully.

Fig. 10: Volume resistance of RR System 5 REFERENCES

The volume resistance is relatively constant until
[1] Küchler A.: Hochspannungstechnik; VDI Verlag 2004, ISBN: 3-
70% r.h. at 9 TOhm and then strong falling. It has to be 540-21411-9
assumed that the volume resistance does not break [2] C. Rupp, “Condition Evaluation of Rotating Electric Machines”,
down, because the bars were tested up to a voltage of 10 Diploma Thesis, Graz University of Technology, 2005
kV AC withstand voltage. For this reason this behaviour [3] Strehl T.: “On- and Off -Line Measurement, Diagnostics and
Monitoring of Partial Discharges on High-Voltage Equipment”,
can be reduced to following observation of the surface Workshop 2000, Alexandria, Virginia, 13 & 14 September 2000,
resistance. Paper No. 4
[4] Ramm G., Moser H.: “Calibration of Electronic Capacitance and
Dissipation Factor Bridges” IEEE Trans. Instrumentation and
Measurement, Vol. 52, No. 2, April 2003