UNSW - SCHOOL OF ELECTRICAL ENGINEERING AND TELECOMMUNICATIONS

ELEC4611 POWER SYSTEM EQUIPMENT LABORATORY

EXPERIMENT 3
NON-DESTRUCTIVE TESTING OF HIGH VOLTAGE COMPONENTS

INTRODUCTION:

The assessment of electrical insulation is an important requirement in the maintenance of

high voltage equipment. The two diagnostic tests which are perhaps the most useful are
the dielectric loss measurement and partial discharge determination.

Dielectric loss (DDF)

The electric field analogue of magnetic hysteresis is important because it can represent a
major heat source in insulation and thus can cause thermal deterioration over time if it is
excessive. Dielectric loss can be measured with a bridge, which incorporates the
insulation as a lossy capacitor. The dielectric loss is given by:
PD =  C V2 tan 

where  is the power frequency, C is the insulation capacitance, V is the applied voltage
and tan is the loss factor of the insulation. For good insulation, the loss factor is small so
tan   and thus dielectric loss is generally expressed by quoting , the dielectric loss
angle (d.l.a.) of the insulation material. The modern terminology is dielectric dissipation
factor (DDF). The most common unit is in milliradians although the modern unit is now
milli-watts per VAR

Partial Discharges (PD)

In the manufacture of insulation, it is inevitable that some gas-filled voids will be left in
the solid material. With an r  1 in these voids, compared to r  2 - 4 in the adjacent
solid material, the electric field in the voids may be high enough to cause some corona
type breakdown in the void. This can then carbonise and enlarge the void, resulting in a
deterioration of insulation quality. Monitoring of the level and number of these 'partial'
(or incomplete) breakdowns gives useful information about the insulation quality. Partial
discharge activity in gaseous insulation is the well-known corona discharge. PD
monitoring provides an indication of the magnitude of discharges (units are in pico-
coulombs) and the repetition rate. These quantities are indicative of the insulation
condition.

Both tests are performed routinely in large utilities and industries on a wide range of high
voltage equipment using commercially manufactured test equipment similar to that used
in the experiments here.

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 1 - 2020

PART A: Measurement of dielectric loss in a 33kV current transformer

Introduction:

The measurement of dielectric losses at high voltage is carried out as a routine or type
test on a number of items used in power systems, e.g. transformers and cables.
Acceptable levels are specified depending on the apparatus and its application. The levels
are characterised by the loss factor tan, with tan   here.

The aim of the test is to ensure that electrical stresses at about operating voltage, and
possibly above, do not produce losses in the dielectric which could result in thermal
failure in service. The usual criterion is that no increase in loss factor (tan) occurs over
the voltage range specified. In addition, absolute values can be specified. Oil impregnated
current transformers might be expected to have values of the order of 3 milliradians for 
at ambient temperatures: the magnitudes can increase if the temperature is raised which
can give useful information. However, the complexity and cost of such tests precludes
their use as routine. Type tests on cables may include such a requirement.

The major cause of a high DDF in new insulation is usually poor drying of the cellulosic
materials. It is normal to achieve 0.1% or better moisture content in paper before
impregnation with dry oil (< 5ppm) under vacuum.

Traditionally, the measurement circuit used is the Schering Bridge which is a 4-terminal
AC bridge with the earthy end of the component under test isolated from earth. The
theory of the circuit is well documented (see Appendix): the main feature is that the
measurement arms are at low voltage and safe for normal operation. However, the
presence of HV fields adjacent to circuits in which very low voltage differences are being
measured requires very careful shielding and arrangements of components. In addition, it
is difficult to eliminate stray 50 Hz effects arising from incorrect earthing and from the
power supply. The detector has a built-in harmonic filter. When balance is achieved, both
tan and the equipment capacitance can be determined from the balanced conditions.

In this experiment, the MI600 universal current measuring system is used to measure the
DDF and capacitance of a 33 kV resin-impregnated current transformer. The MI600 is an
advanced computer-based measuring system with 2 sensor units and a controller. This
instrument functions as an equivalent automatic Schering Bridge.

Apparatus

250 kV testing transformer

150 kV standard gas capacitor, 106.44 pF (reference capacitance)
Omicron Mtronix MI600 Universal Current Measuring System
Sadtem 33 kV resin-impregnated current transformer (test object)
HVAC controller
HV capacitance divider

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 2 - 2020

 Figure 1: MI600 Universal Current Measuring System

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 3 - 2020

Procedure

1. Verify that the circuit is set up as in diagram (Figure 1) and identify each item.

2. Become familiar with the operating software and its features. The system
documentation can be found in the computer. If uncertain, ask for advice.

For all the following steps, use your phone camera to take a photo of every measurement.

3. Raise the applied voltage to 5 kV. Wait for 1 minute and then obtain the DDF and
capacitance readings.

4. Repeat measurement at voltages of 10, 13, 16, 19 and 22 kV (maintain for maximum of
2 minutes). Note that 19 kV is the normal operating voltage of the test object

5. Record values of DDF during voltage reduction, e.g. at 19, 16, 13 and 10 kV.

6. Plot the DDF readings against the applied voltage and comment.

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 4 - 2020

PART B: Partial discharge measurement as a design and diagnostic test -
33 kV current transformer and 11 kV model bushing.

Introduction

The objectives of the experiment are as follows:

(i) To introduce the technique of PD measurement as a practical test.

(ii) To determine PD characteristics of the 33 kV current transformer tested in Part A.

(iii) To show how a change in dielectric medium (oil or air) can increase the voltage level
at which PDs occur in, for example, a model 11 kV bushing.

Apparatus

As in Part A with the following changes:

11 kV model bushing (test object)

Standard capacitor removed
PD blocking capacitor (Ck = 1000 pF)
Omicron Mtronix MPD600 Partial Discharge Measuring System
Discharge calibrator (100 pC, and 1000 pC)
A sharp protrusion (to demonstrate effect of interference from corona discharges)

Circuit

Figure 2: MPD600 Partial Discharge Measuring System

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 5 - 2020

Procedure

1. Disconnect the MI600 system from the circuit, and earth the bottom end of the 33 kV
current transformer. Connect the MPD600 PD measuring system to the test circuit as
shown in Figure 2.

2. Without HV applied, calibrate the measuring system by injecting -50 pC with the PD
calibrator. Remove the calibrator on completion.

For all the following steps, use your phone camera to take a photo of every measurement.

3. Connect a needle-bowl electrode setup to the test circuit and raise the voltage slowly (~
300V/step) until 'corona' discharge can be detected. Record (i) the inception voltage,
i.e. the voltage level at which PDs firstly can be seen on the discharge detector; (ii) the
PD magnitude; and (iii) where the PD activity occurs in relation to the 50 Hz voltage
cycle.

4. Remove the needle-bowl electrode setup and start measuring PDs from the 33 kV CT.
Raise voltage slowly and record (i) the inception voltage; (ii) the PD magnitude; and
(iii) where the PD activity occurs in relation to the 50 Hz voltage cycle. Increase the
voltage and stop at each 3kV step and repeat recording results (i.e. (ii) and (iii) items)
after 1 minute. Repeat measurements up to 22 kV (maintain for maximum of 2
minutes).

5. Reduce the voltage slowly, stop at each 3 kV step and determine (i) the PD magnitude;
and (ii) where the PD activity occurs in relation to the 50 Hz voltage. Estimate the
extinction voltage, i.e. the voltage level at which all PDs vanish.

6. Run down the supply and disconnect the 33 kV CT. Connect the empty 11 kV model
bushing to the HV circuit. Repeat external PD calibration without HV applied.

7. Raise voltage slowly (i.e. 300V/step) until PDs are detected (at about 11kV). Note
voltage and PD value. Raise voltage and hold at each 3kV step (up to 17 kV) and
record PD level plus PD pattern (do not maintain voltage for more than 2 minutes).
Reduce voltage and determine extinction voltage.

8. Add oil to the model bushing and repeat step 7. Limit the maximum voltage to 25 kV
RMS (maintain for maximum of 2 minutes).

9. Reduce voltage to zero and turn off the HVAC controller.

Results and discussion

1. Estimate PD levels in pCs for the different conditions and plot the variation of PD level
against applied voltage for the current transformer.

2. Calculate the electrical stress at the inner conductor surface for air and oil conditions
for the model bushing:

V 1
Ex = .
rx  x r2  x r3
ln + ln
 1 r1  2 r2

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 6 - 2020

 where rx is the radius in medium x at which the stress Ex is required. The calculation
assumes an idealised concentric cylinder system with infinitely long electrodes and no
edge effects at earth rings (which simulate outer clamps or foil).

Figure 3: Dimensions of 11kV bushing.

3. Compare your calculated inception voltage for air with that measured. It is assumed
that discharging occurs at the conductor surface (i.e. maximum stress region). Note
that for uniform gaps, the discharge inception stress is usually taken as 30 kV/cm crest
for a 1 cm gap.

4. Why does the addition of oil increase the discharge inception level?

5. Relate the PD values and DDF (Part A) measurements for the current transformer.
What is the significance?

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 7 - 2020

 APPENDICES
Appendix 1: DDF measurement using the Schering Bridge

The DDF (tan) (and the capacitance C) of a high voltage insulator are normally
measured by a HV AC (4-terminal) bridge technique. The measurement must be done
with the insulation stressed at rated voltage (or higher) to determine its dielectric
condition, because the DDF is voltage dependent. Because of the variation of DDF with
voltage, the test is usually done over a range of voltage above and below the normal
operating value. Usually insulation will exhibit a “turn-up” in the DDF vs. V curve just
above operating voltage. If this insulation is degraded the “turn-up” may occur at a
lower voltage level and this will be an indication of insulation problems.

The most commonly used bridge arrangements for DDF measurements are the Schering
Bridge and the Transformer Ratio Arm Bridge. These have the test object at full
voltage, but the variable balance impedances which may have to be manually adjusted
are effectively at earth potential [modern bridges will be automatically balancing]. Both
of the above types are 4-terminal bridges and thus require both amplitude and phase
balance.

As the Schering Bridge is used in the laboratory experiment, it will be described and
analysed in detail here: the figure below is typical of a Schering Bridge test circuit:

Earth

Z1
Z2
C1 R2
R1
V


H.V.
Z3
C Rx  Cx

Test object
Z4
Figure 4: Schering Bridge balancing circuit

(i) C1 and R1 are the variable impedances (high precision and stable values:
R2 may also be variable).
(ii) R2 is a standard non-inductive wire-wound resistor.
(iii) C is a standard H.V. (150 kV) precision capacitor with negligible losses
(typically100 pF: with gaseous dielectric). It must be able to withstand the same
voltage as the test object.
(iv) Rx and Cx is the test object (represented by a series equivalent combination)

The bridge is energised at the rated voltage of the insulation, but there are only a few
hundred volts at most on R1, C1 and R2 , the components which are manually operated
to achieve balance.

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 8 - 2020

At balance: Z2 = Z1 or Z2 = Z4
Z4 Z3 Z1 Z3

From the circuit diagram, we have:

1 = 1 + jC1
Z1 R1

Z2 = R2
1 = jC
Z3

Z4 = Rx + 1
jCx

and thus, for balance, we have:

R2 1 + jC1 = 1 + Rx jC .
R1 jCx

or: jC1 R2 + R2 = C + jRx C .

R1 Cx
Equating real and imaginary parts, we get:

(a) R2 = C i.e. Cx = R1 C
R1 Cx R2
R2 C1
(b)  R2 C1 =  Rx C i.e. Rx =
C

tan  is the required quantity of the test measurement, and is given by:

tan  = Rx =  Rx Cx =  . R2 C1 . R1 C
Xx C R2

i.e. tan  =  R1 C1 ( is the power frequency value, 100)

Appendix 2: Partial Discharge measurement

A discharging sample may be simply represented by the equivalent circuit in Figure 5.

The applied alternating voltage V is increased until the voltage across the cavity (Cv)
produces a discharge, with a subsequent pulse across the sample of V x .
 CC  (1)
qv =  Cv + c a   Vv
 Cc + Ca 
Cc
where Vv =  Vx
Cc + Cv
and Vx = instantaneous applied voltage

The sudden change Vv will produce a voltage pulse at the sample terminals:

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 9 - 2020

 Cc
Vx =  Vv (2)
Cc + Ca

and the apparent change in charge across the sample on breakdown of Cv is:
 CC 
qx = Vx  Ca + c v   Vx .Ca  Vx .C x (3)
 Cc + Cv 

[Note: Ca also includes any external capacitance that is in parallel with test object, Cx].

If Ca  Cc (this is usually the case but not always, e.g. for a point-plane system) then,
before breakdown of the void, equation (1) becomes:

qv  ( Cv + Cc ) .Vv (4)

and if Cv  Cc [Cv assumed to be much thinner and same cross section as Cc] equation
(3) becomes:
qx  ( Ca + Cc ) .Vx (5)
Cc
= ( Ca + Cc ) .Vv [from eq.2]
Cc + Ca
= Cc .Vv (6)

as measurable at terminals. Note that Cc and Vv are rarely known individually. Now
from (4) and (6):
qv Cv + Cc partial discharge dissipated in void
= =
qx Cc apparent partial discharge at terminal
Cc
or: qx = qv (7)
Cv + Cc
C
 c qv if Cv  Cc
Cv
i.e. the measured PD is much smaller than the actual PD discharged in an
elementary cavity for the simple case considered.

The partial breakdown of the insulation causes a charge reduction and thus creates a
momentary voltage collapse between the two main terminals of the test object. This gives
rise to an electrical current pulse of very short duration and small amplitude. It can be
detected in the external circuit connected to the test object. Note that the small voltage
collapse is superimposed on the large 50Hz supply voltage. Basically, the function of a
discharge detector is to decouple the high-frequency PD signal from the supply and to
amplify this signal.

A conventional circuit to detect the PD current pulse is shown in Figure 6. Cx represents

the capacitance of the test specimen. Ck is the coupling (blocking) capacitor which, of
course, must be discharge-free and appropriately rated for use at high-voltage. The
measuring impedance Z is connected in series with Ck. Following a PD, a redistribution
of the remaining charge between Cx and Ck will give rise to a corresponding current pulse
flowing through Ck and the measuring impedance Z. The signal picked up by Z is fed
through a long cable to an amplifier which is at some metres away from the HV area. In
some systems, the amplifier is incorporated into Z. Measurement sensitivity is dependent

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 10 - 2020

on the relative values between Cx and Ck as this will determine the proportion of restoring
charge flowing through Z. Preferably, Ck should be as large as possible to maximise the
external charge displacement.

Figure 5: (a) Representation of a void in a solid, (b) the

equivalent circuit, and (c) possible voltage waveforms.

Figure 6: Basic circuit for PD measurements.

====================

ELEC4611 - Experiment 3: Non-destructive testing of HV components - Page 11 - 2020