Field diagnostics and diagnostic monitoring

of HV insulation at generators, motors,

transformers and cables (part I)

March 5, 2015

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 22

Introduction
Meta-Trends

1. HV asset fleet / installed base becomes older

2. Knowledge is a limited resource
3. HV assets are designed with lower margins than the generation
before
4. Cost pressure

OMICRON

Page 3

but on the other hand ...

March 10,
2015

Page: 4

Blackouts
millions of
location
people affected

date

July 2012 India blackout

620

India

30 July 2012-31 July 2012

January 2001 India blackout

230

India

2 January 2001

November 2014 Bangladesh blackout

150

Bangladesh

1 November 2014

2015 Pakistan blackout

140

Pakistan

26 January 2015

2005 JavaBali blackout

100

Indonesia

18 Aug 2005

1999 Southern Brazil blackout

97

Brazil

11 March 1999

2009 Brazil and Paraguay blackout

87

Brazil, Paraguay

1011 Nov 2009

Northeast blackout of 2003

55

United States, Canada

1415 Aug 2003

2003 Italy blackout

55

Italy, Switzerland, Austria,

Slovenia, Croatia

28 Sep 2003

Thailand Nation-wide blackout of 1978

40

Thailand

18 Mar 1978

Northeast blackout of 1965

30

United States, Canada

9 Nov 1965

March 10,
2015

Page: 5

What to do ?

maintenance

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 77

Historical stages of maintenance practices

No maintenance (run to failure)

Preventive (time based)
Predictive (condition based)

Insulation strength

Condition and Time based maintenance

New condition

Condition based maintenance

Diagnostic level
Time based maintenance
Operating stress
Time

Condition based maintenance

requires reliable diagnostic tools

Equipment, that can be

easily replaced within a
month

Run to Failure

Non critical equipment,

that can be replaced
within few months

Preventive
Maintenance

Critical equipment that

should be repaired or
replaced within 1 year

Predictive
Maintenance

Critical Equipment with

long term delivery time
(>1 year)

Real Time
Monitoring

Cost for unforeseen shut down

Ease of repair/replacement

Reliability Centred Maintenance

Reliability Centred Maintenance

Reliability Centred Maintenance

Reliability
cantered
Maintenance
RCM

Preventive
Maintenance

PM

Condition
Based
Maintenance

Run to Failure

CBM

Real Time
Monitoring

RTM

RTF

Predictive
Maintenance

PdM

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 13
13

Measurement vs. Monitoring

Measurement is a process of making a quantitative assessment of an
unknown quantity.
Monitoring is the systematic observation of repeated processes or
systems in order to draw conclusions through the comparison of results.
The comparison of repeated measurements is defacto monitoring.
Challenges:
Boundary conditions, comparability, measurement uncertainty,
repeatability

OMICRON

Page 14

Measurement vs. Monitoring

Why conduct monitoring when design, type, series and commissioning
tests are performed?
> Design, type and serial tests are conducted prior to asset installation
> Possible damage via transport, assembly and installation that could
lead to eventual failure
> These tests do not assess online condition when asset is in service
A commissioning test is:
> Limited in time
> Often limited to specific stress factors, but not all
> A snapshot in time, not continuous
Example:
Cable after laying test: 1h @ 1.1 Un, no load, no temperature stress

OMICRON

Page 15

Measurement vs. Monitoring

Monitoring enables measurement comparisons to be made in
the presence of all operational stress factors (3-phase ,
asymmetries, couplings, loads, vibrations, temperatures, etc..)
during real operation conditions over longer periods of time.

Monitoring helps to predict severe failure to extend reliable

operation of equipment throughout its service life.

OMICRON

Page 16

Measurement vs. Monitoring

OMICRON

Page 17

Measurement vs. Monitoring

> Periodic offline diagnostic measurements
>
>
>
>

Allows the variation of different measurement parameters

Does not take place under normal operating conditions
Execution is highly dependent on human interaction
Provides only a snapshot of condition state

> Periodic online monitoring (temporary monitoring)

>
>
>
>

Does not allow the variation of different measurement parameters

Takes place under operating conditions (load, temperature, vibration)
Execution required a minor degree of human interaction
Provides a snapshot of condition state over a specified period of time

> Permanent monitoring (continuous monitoring)

>
>
>
>

OMICRON

Does not allow the variation of different measurement parameters

Takes place under operating conditions (load, temperature, vibration)
Execution does not require human interaction after setup
Provides continuous (on-going) condition assessment

Page 18

Diagnostic monitoring
> Continuous or periodic monitoring of HV equipment is an essential tool for
effective maintenance management
> Monitoring answers questions about the present condition of the
equipment and its future performance
> The large amount of real-time data gathered by monitoring systems can
be used for precise insulation condition assessment
> Effective comparison of historical data is enabled via an easy-to-use and
extendable database solution
> Exact knowledge of insulation state at any time saves money, as
maintenance schedules can be specifically optimized and the service life
of assets can be extended

OMICRON

Page 19

Diagnostic monitoring
> Involves no operational variables
> Provides information to evaluate the aging condition of equipment
> Allows assessment of future developments in the aging process
> Supports decision making for ongoing maintenance or the replacement
of components
> is a "planning tool (asset management)

OMICRON

Page 20

What we would like to discuss today...

OMICRON

Page 21

What we would like to discuss today...

aging of HV assets means in general aging of HV
insulation systems...
therefore we like to discuss today
> measurement methods / diagnostically methods to determine
health indicating parameters of HV insulations
> monitoring of diagnostically parameters YES
> monitoring of operational parameter NO

OMICRON

Page 22

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 23
23

Field Diagnostics & diagnostic monitoring of rotating machines

Copyright "Siemens Pressebild

OMICRON

Seite 24

Field Diagnostics & diagnostic monitoring of rotating machines

OMICRON

Seite 25

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 26
26

Life Time expectation

Opinions & Statements:

> 20 ... 30 years (till rewinding) ?

> economically 15 years (depreciation) ?
> operated at determined temperature limit 20.000 hours (= 2.3 years)

all above may be correct what we learn: prediction is difficult ... impossible,
therefore diagnostically methods or monitoring is needed

OMICRON

Seite 27

Life Time expectation

Insulation design evolution the last 100 years

Glew C.N.:The Next Generation A Review of the Factors influencing the Output
of an Electrical Machine in the New Millenium., INSUCON/ISOTEC 98, p. 231-242
OMICRON

Seite 28

Generator

OMICRON

Seite 29

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 30
30

Faults & Aging

Faults

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

CIGRE 2009 / Survey of Hydro Generator Failures

OMICRON

Seite 31

Faults & Aging

Insulation of Stator Winding general
Coil Type Winding

OMICRON

Roebel Bar Type winding

Seite 32

Faults & Aging

Insulation of Stator Winding general

OMICRON

Seite 33

Faults & Aging

Stator slot

OMICRON

Seite 34

Faults & Aging

Insulation of Stator Winding general
A.
B.
C.
D.

Covering Tape
Spacer, Coil-End Bracin
Groundwall Insulation, Mica Tapes
Grading/ Silicon Carbide Coating

E.
F.
G.
H.
I.

Slot Semi conductive Coating

Inner Semicon. coating
Turn Insulation
Slot Wedge / Seal
Stator Core

OMICRON

Seite 35

Faults & Aging

Insulation of Stator Winding general

End winding Grading potential grading,

silicon carbide, extends only few
centimetre beyond the end of each slot.

Slot Semiconductive Coating prevent PD

from occurring in any air gap that might be
present between the coil/ bar surface and
the stator core. Carbon-loaded paint or tape

OMICRON

Seite 36

Faults & Aging

Insulation of Stator Winding general
Standards
IEC 60034

Rotating electrical machines

IEC 60034-18

Functional evaluation of insulating systems

IEC 60034-18-31 Test procedures for form-wound windings Thermal evaluation and classification of
insulating systems used on machines up to and including 50 MVA and 15 kV
IEC 60034-18-33 Test procedures for form-wound windings Multifactor functional evaluation; Endurance
under combined thermal and electrical stresses of insulating systems used in machines
up to and including 50 MVA and 15 kV
IEC 60034-18-34 Test procedures for form-wound windings - Thermomechanical cycle endurance evaluation
IEC 60085

Thermal evaluation and classification of electrical insulation

IEC 60216

Guide for the determination of thermal endurance properties electrical insulating

materials

IEC 60371

Specification for insulating materials based on mica

IEC 60270

High-voltage test techniques Partial discharge measurement, Version 2000, 3rd Edition

IEC 60034-27

TS, Ed.1, Rotating electrical machines Part 27: Off-line partial discharge measurements
on the stator winding insulation of rotating electrical machines, Version 2006

IEEE 1434-2000 IEEE Trial-Use Guide to the Measurement of Partial Discharges in Rotating Machinery,
Version 2000
OMICRON

Seite 37

Faults & Aging

Broken Solder Connection

OMICRON

Seite 38

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Insulation with Cavity

OMICRON
Seite 39

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Discharge Propagation Path

OMICRON
Seite 40

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Loose coils in the slot, semi conductive coating abrasion

OMICRON
Seite 41

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Delamination within Mica-Insulation

OMICRON
Seite 42

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Delamination within Mica-Insulation

OMICRON
Seite 43

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

40 years old Insulation

OMICRON
Seite 44

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Semi conductive/ grading coating overlap failure

OMICRON
Seite 45

Source: Brtsch et al. "Insulation Failure Mechanisms of Power Generators", DEIS July/August 2008

Faults & Aging

Partial Discharge at the End-Winding

OMICRON
Seite 46

Faults & Aging

Source: VATECH, AUSTRIA

Partial Discharge at the End-Winding

OMICRON

Seite 47

Faults & Aging

Insulation faults are mainly caused by aging and overloads
thermal
> overload
> extreme high and low ambient temperatures
> Cooling air deficiency / wrong installation height above sea level; Management
errors for large generators (iron and copper different expansion coefficients)

ambient (chemical)
> Moisture + (water with nitrogen oxides under PD influence generates nitric acid)
> Oxidation by aggressive environment

electrical
> Overvoltages, transients or permanent stress

mechanically
>
>
>
>
>

OMICRON

pollution
misalignment
Vibrations / relaxations
Exposure by foreign object damage
Forces caused by short-circuits & faulty synchronization actions

Seite 48

Faults & Aging

OMICRON

Highest and lowest temperature

Increase of Current through Overload

Rated Voltage
Test Voltage
Over Voltage

water, ice
aggressive and reactive chemicals
abrasive particles: metal parts, ash,
carbon, lubricants

Oscillation in slot section

mechanical forces in enwinding section
Different thermal expansion

Seite 49

Topics
> Introduction
> Maintenance
> Measurement, Diagnostics and Diagnostic Monitoring
> Rotating Machines (Generators / Motors)
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities
Insulation Resistance
Polarization Index
C / tan (power factor)
Partial discharge

Page
Page 50
50

Life Time expectation & Aging

Diagnosis and Monitoring can not avoid faults, but it helps to plan and
minimize the cost
A suddenly popped up insulation damage is not repaired on average less than
6 to 8 weeks (open end) - if all goes perfect.
A well-planned repair of a machine takes 2-3 weeks till re-commissioning
(depending on the size of the machine)

OMICRON

Seite 51

Measurement & monitoring parameters /quantities

> Visual inspection
> Insulation Resistance
> Polarization Index
> Partial Discharge
> Capacitance
> TanDelta / Power Factor Tip Up

OMICRON

Seite 52

Measurement & monitoring parameters /quantities

Insulation Resistance / Polarization Index
Insulation Resistance (IR) and Polarization Index (PI) are two universally accepted
diagnostic tests for insulation tests. These have been in use for more than 75 years.
The IR test measures the resistance of the electrical insulation between the copper
conductors and the core of the stator or rotor. Ideally the value of this resistance is
infinite since the purpose of the insulation is to block current flow between the copper
and the core. But in practice, it is not possible.
However, the resistance should have a high value to avoid any appreciable leakage
current. Lower value of IR indicates that the insulation has deteriorated.

OMICRON

Seite 53

Measurement & monitoring parameters /quantities

Insulation Resistance /
Polarization Index

OMICRON

Source: [Link]

In insulation resistance test, a high DC

voltage is applied across, conductor and
ground. The voltage is applied across the
insulator. Due to this applied high DC
voltage there will be a current through the
electrical insulator. Insulators are
dielectric in nature (capacitance). Due to
that, initially there will be a charging
current. After some time when the
insulator is totally charged, the capacitive
changing current becomes zero and then
only resistive conductive current presents
in the insulator. That is why it is always
recommended to do insulation resistance
test at least for 1 minute as it is proved
that charging current totally becomes
zero after 1 minute.

Seite 54

Measurement & monitoring parameters /quantities

Insulation Resistance / Polarization Index
Effect of Temperature on IR
Unfortunately, just measuring IR has proved to be
unreliable, since it is not tenable over time and strongly
dependent on temperature. A 10C increase in temperature
can reduce IR by 2 to 10 times. When readings of
temperature and insulation resistance are plotted on
ordinary equally divided co-ordination, a curved
characteristics is obtained. On the other hand if graph
paper is used on which the insulation scale is laid out in
logarithmic division, the graph becomes a straight line.
Further, the effect of temperature is different for each
insulation material and type of contamination. Although
some temperature correction graphs and formulae are
given in the IEEE-43 and some other books, they are
acknowledged as being unreliable for extrapolation by more
than 10C. The result is that every time IR is measured at
different temperatures, one gets a completely different IR.
This makes it impossible to define a scientifically
acceptable IR value over a wide range of temperatures

OMICRON

Seite 55

Measurement & monitoring parameters /quantities

Insulation Resistance / Polarization Index
Effect of Temperature on IR
Unfortunately, just measuring IR has proved to be
unreliable, since it is not tenable over time and strongly
dependent on temperature. A 10C increase in temperature
can reduce IR by 2 to 10 times. When readings of
temperature and insulation resistance are plotted on
ordinary equally divided co-ordination, a curved
characteristics is obtained. On the other hand if graph
paper is used on which the insulation scale is laid out in
logarithmic division, the graph becomes a straight line.
Further, the effect of temperature is different for each
insulation material and type of contamination. Although
some temperature correction graphs and formulae are
given in the IEEE-43 and some other books, they are
acknowledged as being unreliable for extrapolation by more
than 10C. The result is that every time IR is measured at
different temperatures, one gets a completely different IR.
This makes it impossible to define a scientifically
acceptable IR value over a wide range of temperatures

OMICRON

Seite 56

Measurement & monitoring parameters /quantities

Source: [Link]

Insulation Resistance / Polarization Index

OMICRON

Seite 57

Measurement & monitoring parameters /quantities

Insulation Resistance / Polarization Index
PI is a variation of the IR test. It is the ratio of IR measured after voltage has been applied for 10
minutes (R10) to the IR measured after one minute (R1), i.e.
PI = R10 / R1
PI was developed to make interpretation of results less sensitive to temperature. PI is the ratio of two
IR at two different times. Temperature of the winding does not rise during the test period of 10
minutes. So it is fairly assumed that both R10 and R1 are measured at same winding temperature.
Then the temperature correction factor will be same for both cases and will be cancelled during the
calculation of Pl. Thus PI is relatively insensitive to temperature.
Interpretation of Polarisation Index results
PI

Condition of item under test

<1.0

Hazardous

1 - 1.5

Bad

1.5 - 2.0

Doubtful

2.0 - 3.0

Adequate

3.0- 4.0

Good

> 4.0

Excellent

OMICRON

Seite 58

Measurement & monitoring parameters /quantities

Insulation Resistance / Polarization Index
U-VWE
Measurement
U-VWE

60 s
Time

Current

600s

3.7637 nA

PI = 25.98 / 3.76 = 6.9

OMICRON

Seite 59

25.9809 nA

Polarization and frequency

> Due to different polarization methods the dielectric constant (r) and the
dissipation factor (tan()) are frequency dependent
dielectric r
constant

electrical
optics
engineering
boundary layer polarization

refraction index n

r = n

orientation polarization
grid polarization
atom polarization
1
0

Hz

polarization
losses tan()

kHz

MHz

wavelength

GHz
m

frequency

IR
losses caused
by reloading of
partial
capacitances

thermal losses

dipoles follow
the field with
delay

mm
light

grid gets into

resonance

nm
UV

x-rays
atoms get
excited

light absorption (optics)

Conductivity

> Caused by current flowing through the

insolation
> Mostly caused by moving electrons
> In liquids also caused by ion movement
> In most insolation materials the water
concentration determines the conductivity

> Surface currents should be suppressed or

relatively small
> Causes losses at all measuring frequencies
> Determines the losses at very low frequencies,
below 0.1Hz

insulation
surface
current

current through
insolation

Partial discharge (PD) losses

> PD in the Insulation can cause additional
losses
> PD act as a current though the insulation

> PD losses only occur above the inception

voltage
> With a higher voltage PD losses are higher due
to higher PD intensity
> A precise PD detection is not possible
> Only high PD activity leads to a visible increase
of the tangent()
> This mostly happens far above the inception
voltage
> PD sensitive insulations can get seriously
damaged at this PD level (e.g. PE)
> Use a PD measurement system for PD
detection (e.g. the Omicron MPD600)

insulation
cavity
with PD

Vacuum capacity
> Capacity of a test object which has as dielectric vacuum

> For the most objects it is only an ideal concept

> Has no losses

vacuum

> Dielectric constant is 1

> Gas insolation is close to the vacuum insolation

Polarization, conductivity and vacuum capacity

> The electric representation of an
insolation consists of:
> An ideal capacity for the vacuum
field
> Different polarizations, here
shown as one
> A electric conduction through the
dielectric

> This can be modeled as an

equivalent circuit

vacuum
E-field

polarization

conductivity

equivalent circuit

Image Source: Andreas Kchler, Hochspannungstechnik, Heidelberg, 2009

Equivalent circuit
> The equivalent circuit can be
summarized to two elements:
> A capacitor
> A resistor
> This gives the equivalent circuit
for the tangent() measurement
> As parallel circuit
> A series circuit is also possible

Image source Top Image: Andreas Kchler, Hochspannungstechnik, Heidelberg, 2009

What is the dissipation/loss factor tangent()?

> When using the equivalent circuit diagram of the test
object, the tangent() is the relation between the
resistive current (IR) and the capacitive current (IC)
> 

> But IR and IC are not directly measureable

IC
test
object

IR
Itest

What is the dissipation factor tangent()?

U

> In the complex plane the angle delta can be

seen between the test object current (Itest) and
the capacitive current (IC)
> Using sinusoidal currents this gives a phase
difference between the test object current (Itest)
and the capacitive current (IC)

i(t)

IR
IC

= phase difference
test object
current

capacitive
current

Itest

Difficulties measuring tangent()

> A good ground connection is essential
> Low inductance and resistance
> Good connection to the grounding point
> Star shaped grounding of all used devices

> Use short connection cables

> Very good insolation against stray currents needed
> Test object isolation to the ground (best use PTFE or PE)
> High voltage to test object (use guard electrode if possible)

> All contacts in the measuring circuit have to be as good as possible

Typical tangent() values at 50Hz for different

insolation materials

Insolation material

Typical tangent()
(50Hz)*

Polyethylene (PE and XLPE)

8*10-5 - 4*10-4

PTFE / Teflon

~1*10-4

Polyvinyl chloride (PVC)

~2*10-2

Cast resin (filling and humidity dependent)

5*10-3 - 2*10-1

Oil impregnated paper (0,1%-10% humidity)

2*10-3 - 1

Mineral oil (humidity dependent)

1*10-3 - 8*10-2

Silicone oil

~2*10-4

Mica

Lower than 1*10-3

Glass

~1*10-4

Pressboard

~3*10-3

* Source: Andreas Kchler - Hochspannungstechnik

Dielectric losses, conductivity and temperature

> In most insulation materials the conductivity increases with the temperature
> This gives a higher dissipation factor
> Under certain conditions this can lead to a thermal break down

> The losses due to polarization may sink with higher temperatures due to
better dipole movement
> This gives a lower dissipation factor

> In combination this mostly gives a rising dissipation factor with rising
temperature, but other results may also occur

Measurement & monitoring parameters /quantities

TanDelta

tan G

Dissipation Factor

cos I
I

Insulation

tan G

OMICRON

cos M

Seite 71

Power Factor

P
(Dissipation Factor DF)
Q
P
(Power Factor PF)
S

Measurement & monitoring parameters /quantities

Dielectric Losses

.....

Conductive Losses

OMICRON

Polarization Losses

Seite 72

Measurement & monitoring parameters /quantities

Dielectric Losses
> Conductive losses
> Movement of conductive particles
> carbon in oil

> Movement of ions and electrons

> leakage current through the insulation or on the surface of a bushing
> Partial discharge

> Polarization losses

> Interfacial polarization
> Polarization of dipoles in insulation material (rotation/suspension)
> Increased moisture causes more dipoles and hence more losses

OMICRON

Seite 73

Measurement & monitoring parameters /quantities

tanDelta = f(f)

tan
Sum

Polarization Losses

Conductive Losses

f [Hz]

OMICRON

Seite 74

Measurement & monitoring parameters /quantities

tanDelta = f(f)

50Hz

5.000
DF
2.000
1.000
0.500
0.200

0.12

0.100

Moderate

New

Aged

0.050
0.020
0.010

0.0036

0.005

0.0024
0.0001

OMICRON

0.001

0.010

0.10

1.0

10.0

Seite 75

Freq/Hz 1000

Measuring Principle

Reference path: tan G = 10-5

25 mm
2.5 km

Measurement & monitoring parameters /quantities

Capacitance and TanD

tan G =
conductor (tan GC)
polarization (tan GP)
losses caused by PD (tan GPD)

tan G = tan GC + tan GP + tan GPD

tan G = f (U)

Tan-Delta Test IEC 60894

Tip-Up Test IEEE 286

OMICRON

Seite 77

Measurement & monitoring parameters /quantities

Power factor / dissipation factor (tan delta)
tip-up test at 50 Hz / 60 Hz

Aged rotating machine

Aged rotating machine

Delta tan delta

Power factor / dissipation

factor (tan delta)

Determination of stator winding condition typical curves

for aged and new rotating machines

New rotating machine

New rotating machine

V

The tip-up test and the delta tan delta diagram show the same thing.
The delta tan delta diagram is frequency independent.

OMICRON

Seite 78

Measurement & monitoring parameters /quantities

Power factor / dissipation factor (f) at 2 kV

OMICRON

Seite 79

Power factor / dissipation factor (tan delta)

Power factor / dissipation factor (tan delta)

Power factor / dissipation factor (tan delta)

tip-up test with variable frequency

Power factor / dissipation factor (V) at 62 Hz

What is partial discharge ?

> Partial discharge (PD) is a localized dielectric breakdown of a small portion
of a solid or liquid electrical insulation system under high voltage stress.

> Definition from IEC 60270 Specification:

Localized electrical discharge that only partially bridges the insulation
between conductors and which can or cannot occur adjacent to a conductor.

Important units

electron

Charge Q [1Coulomb] = [1As]:

2nd orbit

Charge in movement = CURRENT!!!

nucleus

The more electrons per TIME interval, the higher is the current
Current

Charge

per

Time

[A]

[Coulomb]

per

[s]

[A]

[Coulomb]

[s]

Charge

Current

times

Time

Charge

Current

Time

q = 

1st orbit

Charge How to measure?

> Time Domain Integration

q = 

SI prefixes

Most used prefixes:

kilo Volts

[kV]

milli Ampere [mA]

micro Farad [F]
nano Coulomb [nC]
pico Coulomb [pC]

Types of PD
> Internal PD
> Void discharges, electrical treeing

> External PD
> Corona
> Surface discharges

Outer semicon

Solid
insulation
Inner
semicon
Image Source: Andreas Kchler Hochspannungstechnik

Solid insulation

External PD

Surface discharge

Corona discharge

Internal PD

Internal discharge
in laminated material

Cavity / void discharge

Treeing

PD classification

En = 2

Semicon layer protrusion

(stress concentration at the tip)

Lines of
electrical field

Void
(field strength doubling)

Breakdown theory in solid materials

1. Intrinsic breakdown (pure electrical breakdown)
> emerges in insulations stressed temporarily with surge voltage

2. Thermal breakdown
> thermal instability of the insulation caused by high temperature

3. Partial discharge breakdown (erosion breakdown)

> as a result of aging processes in electrical high stressed insulation areas
> mostly caused by manufacturing

Main differences in comparison to breakdowns in air

> Failure is destroyed by the breakdown channel
> Breakdown in solid insulations are dependent of:
> purity of different solid materials (e.g. compounding of cables)
> technological influences during manufacturing and contacting
> surface and boundary issues
> temperature, humidity and pressure inside the solid

Thermal breakdown
> dielectric losses are caused by conductivity and polarization phenomena
> material heats up (conductivity raises exponential with temperature)

> higher losses

tan()
polarization losses

conductivity losses

Electrical insulations
are at the same time
thermal insulations

Thermal breakdown schematic process

Dielectric losses cause a warming in the insulation
> highest temperature occurs in the center plane

> not enough heat can be conducted

> heating effects increase

> thermal hot spot

hot spot

Thermal breakdown schematic process

> conductivity rises in heated area

> increasing e-field at the channel ends

> growth of channel to the electrodes

> breakdown

Partial discharge breakdown

> PD occurs as a consequence of :
> local heating (creation of voids)
> abrasion
> delamination
> mechanical stress (vibration)
> defects in the material
> water treeing
> high electric field strength

Partial discharge breakdown (treeing)

> The thinner the insulation, the faster follows on a channel growing a
breakdown

A larger thickness of the insulation

increases the treeing time
> The channel building time is much higher than the channel starting time.

PD sources in solid insulations

cracks cleaving

voids

cavity

delamination of fibers

PD treeing
Acrylic glass

Epoxy resin

PD treeing stopped

AC frequency influences the electrical treeing

Prof. Dr. Ing. Daniel Pepper, Dissertation, Berlin, 2004

Partial discharge breakdown (treeing)

> PD tree reacts as pushed forward potential

> PD treeing is not a steady process:

1. treeing in a channel ends because of

high pressure due to products of
decomposition
Eg

2. high conductivity delays the growth of

trees
++
3. free space charges lower the electric
field locally

Eg: global electric field

How to analyze PD
> Well known approaches:
> PRPD
> Trend
> Q(U)
> PSA (Pulse Sequence Analysis)
> TDR and STDR (PD localization)

> OMICRONs new approaches:

> Frequency based pulse discrimination (3CFRD/3FREQ)
> Cross-talk evaluation (3PARD)

Phase resolved partial discharge PRPD

PRPD correlation between PD pulses and voltage phase
PD nature might be identified

Phase resolved partial discharge PRPD

Trigger

Trigger

Trigger

Trigger

Trigger

Trigger

50
...
Time

Amplitude

Amplitude

> How is a PRPD created?

Time (ms)
10
20

Further ways of PD analysis Trend

Charge vs. time

Applied voltage vs. time

PD repetition rate vs. time

PD repetition rate vs. applied voltage

Further ways of PD analysis Q(U)

> Charge over Test Voltage

Further ways of PD analysis 3CFRD (3 Center Frequency Ratio Diagram)

> Pulse Shape Analysis: 3CFRD or Time/Frequency map

T/W map

Influence of inverse gating on external disturbances

External disturbance

PD

Noise

Further ways of PD analysis 3PARD

> Analysis of synchronous signal cross talk by using the
3 Phase Amplitude Ratio Diagram
500pC

850pC

900pC

Further ways of PD analysis 3PARD

> Analysis of signal cross talk

500pC

Center Phase

850pC

900pC

Right Phase
Left Phase

Further ways of PD analysis 3PARD

Noise

PD source 1

PD source 2

Further ways of PD analysis PD location

> TDR: Time Domain
Reflectometry

> LOC: Statistical TDR

> Dual End:

Measure 2 PD pulse components directly without using reflections

Corona discharges on high voltage potential

Corona discharges on ground potential

Corona discharges on ground potential with multiple

reflections (HV cable)

origin

reflection

Corona during online measurement

Partial discharge mechanism Void

Ut(t)
U1(t)

tt

Void discharges in solid dielectric material

some single cavities

Void discharges in solid dielectric material

Bigger cavities

Some single voids

Void discharges in XLPE cable Early stage (1/3)

Void discharges in XLPE cable Developing fault (2/3)

Void discharges in XLPE cable Developing fault (3/3)

Cable PD Defective outer semicon layer

Bubble in oil

Pattern is disappearing and re-appearing from time to time.

Spike on ground potential in oil

Surface discharges

Surface discharges Significant differences in amplitude

Contact PD

When the bad contact becomes better, the contact PD can disappear.

Floating potential

Delamination on outer semicon layer

Hopping particle

Hopping particles

Hopping particles

Bushing (RBP) Early stages (88kV)

Bushing (RBP) Early stages (126kV)

Bushing (RBP) 156kV

Voltage transformer Inner PD

PD patterns and classification

Source: J. Fuhr, Procedure for Identification and Localization of PD, IEEE Transactions 2005

PD patterns and classification

Source: J. Fuhr, Procedure for Identification and Localization of PD, IEEE Transactions 2005

PD patterns and classification

CIGRE WG 21.03
Recognition of Discharges

Inception / extinction voltage

PDEV

PDIV

Specified threshold
for PDIV and PDEV
noise level
U/Umax
PD magnitude as a function of the normalized test voltage Qm = f (U/Umax)

Full bandwidth

oscilloscope signal time domain o

m bar indicator frequency domain

Low-pass

High-pass

Band-pass
> key element of the PD
measuring system

Time domain vs frequency domain power quality

Time domain vs frequency domain

Time domain vs frequency domain

> easier to display

50 Hz: 1 V
500 Hz: 0.1 V

Time Domain vs Frequency Domain

50 Hz: 1 V
150 Hz: 0.3 V

Time domain vs frequency domain

50 Hz:
150 Hz:
250 Hz:
350 Hz:
450 Hz:

1V
0.3 V
0.1 V
0.03 V
0.01 V

Time domain vs frequency domain

> the dirac pulse consists of all

thinkable frequency components

N x 50 Hz:

1V

The DIRAC pulse

amplitude (A)

time (t)

The area below the curve is 1.

frequency (f)

A (f) = 1, for all f

Time domain vs frequency domain: i(t) and A(f) and

charge
i

area = q = 10pC
area = q = 10pC
area = q = 10pC
area = q = 10pC

t
cut-off frequency

all spectral curves show the

same amplitude

charge = q =
10pC

Time domain vs frequency domain: i(t) and A(f) and

charge

The spectral amplitude

@ 0Hz (f< fcut-off)

represents the impulse

charge q!

frequency f = 0Hz

charge = q =
10pC

PD pulse: time domain vs frequency domain

a)

b)

i (t)

1,5
3

imax
1

imax/2

F()

0,5
2

F(0)
T1

T2
ideal PD-impulse:

0
103 104 105 106 107 Hz

a) time domain display of current i(t)

T1 time to the maximum of the current (imax)
T2 time to the declining half of imax
b) normalized amplitude density F()/F(0)
1) T1/T2 = 1 s/5 s
2) T1/T2 = 5 ns/50 ns
3) T1/T2 = 5 ns/15 ns

109

Partial discharge breakdown

> PD occurs as a consequence of :
> local heating (creation of voids)
> abrasion
> delamination
> mechanical stress (vibration)
> defects in the material
> water treeing
> high electric field strength

This one...

Introduction
> ...For many years, the measurement of PD has been employed as a
sensitive means of assessing the quality ... and localize sources of PD in
used electrical winding insulation ...

> ... Compared with other dielectric tests (i.e. the measurement of dissipation
factor or insulation resistance) the differentiating character of partial
discharge measurements allows localized weak points of the insulation
system to be identified...

Introduction
> ...The PD testing ... is also used when inspecting the quality of new
assembled ... stator windings, new ... components (e.g. form-wound coils
and bars, HV bushings, etc..) and fully impregnated stators...

> ...PD can provide information on

> points of weakness in the insulation system
> ageing processes
> further measures and intervals between overhauls

Introduction
> Partial discharge testing of stator windings can be divided into two broad
groups:
a. Off-line measurements, in which the stator winding is isolated from the power
system and a separate power supply is employed to energize the winding;
b. On-line measurements, in which the rotating machine is operating normally and
connected to the power system.
Annex A
and
IEC 60034-27-2

> Both of these approaches have advantages and disadvantages...

Introduction LIMITATIONS
> ... different types of PD measuring instruments will inevitably produce
different results ...

> ... PD measurements will only be comparable under certain conditions ...

> ... absolute limits for the windings of rotating machines, for example as
acceptance criteria for production or operation, are difficult to define ...
>
>
>
>
>

Pulse propagation phenomena

Difficulties with calibration
Individual frequency response characteristics
Type of PD source
Location within the stator

Introduction LIMITATIONS
> ... Empirical limits verified in practice can be used as a basis for evaluating
test results ...

> ... PD trend evaluation and comparisons with machines of similar design
and similar insulation system measured under similar conditions, using the
same measurement equipment

> ... not all insulation-related problems in stator windings can be detected
by measuring PD (e.g. insulation failures involving continuous leakage
currents due to conductive paths between different elements of the
insulation or pulseless discharge phenomena) ...

1 Scope
> Measuring techniques and instruments,
> The arrangement of test circuits,
> Normalization and testing procedures,
> Noise reduction,
> The documentation of test results,
> The interpretation of test results

> Test with AC <= 400Hz

> ... bars or form wound coils with (without) conductive slot coating...
> Voltage rating >= 6kV

2 Normative references
> IEC 60060-1, High-voltage test techniques Part 1: General definitions and
test requirements

> IEC 60060-2, High-voltage test techniques Part 2: Measuring systems

> IEC 60270:2000, Partial discharge measurements

3 Definitions
> ... general terms and definitions for partial discharge measurements given
in IEC 60270 apply...

> Offline measurement

> Measurement taken with the rotating machine at standstill, the machine being
disconnected from the power system
> Test voltage is ... from a separate voltage source.

> Online measurement

> Measurement taken with the rotating machine in normal operation

3 Definitions (Coating)
> Stress control coating
> Paint or tape on the surface of the groundwall insulation that extends beyond
the conductive slot portion coating in high-voltage stator bars and coils
> The stress control coating reduces the electric field stress along the winding
overhang to below a critical value that would initiate PD on the surface. The
stress control coating overlaps the conductive slot portion coating to provide
electrical contact between them.

> Conductive slot coating

> Conductive paint or tape layer in intimate contact with the groundwall insulation
in the slot portion of the coil side, often called semiconductive coating
> This coating provides good electrical contact to the stator core.

3 Definitions (PD types)

> Slot discharges
> Discharges that occur between the outer surface of the slot
portion of a coil or bar and the grounded core laminations

> Internal discharges

> Discharges that occur within the insulation system

> Surface discharges

> Discharges that occur on the surface of the insulation or on
the surface of winding components in the winding overhang
or the active part of the machine winding

3 Definitions (Evaluation)
> Pulse height distribution
> The number of pulses within a series of equally-spaced windows of pulse
magnitude during a predefined measuring time

> Pulse phase distribution

> The number of pulses within a series of equally-spaced windows of phase
during a predefined measuring time

> Partial discharge pattern

> PD distribution map of PD magnitude vs a.c. cycle phase position, for
visualization of the PD behaviour during a predefined measuring time, in which
specific PD parameters are used for graphical representation

3 Definitions (Evaluation)
> Pulse height distribution

> Pulse phase distribution

> Partial discharge pattern

3 Definitions (Hardware)
> Coupling device
> Usually an active or passive four-terminal network that
converts the input currents to output voltage signals
> These signals are transmitted to the measuring instrument by a transmission
system. The frequency response of the coupling device is normally chosen at
least so as to efficiently prevent the test voltage frequency and its harmonics
from reaching the measuring instrument.

> PD coupling unit

> A high voltage coupling capacitor
of low inductance design and a
low voltage coupling device in series

3 Definitions (Numeric measures)

> Largest repeatedly occurring PD magnitude Qm
> The largest magnitude recorded by a measuring system which has the pulse
train response in accordance with 4.3.3 of IEC 60270, or the magnitude
associated with a PD pulse repetition rate of 10 pulses per second (pps), which
can be directly inferred from a pulse height distribution.

3 Definitions (Numeric measures)

> Normalized quantity number NQN
> Normalized area under a straight line fitted to the pulse counts in each
magnitude window of a pulse height analysis, in which the pulse counts are
expressed as a logarithm of the pulses per second and the pulse magnitude
window is a linear scale.

4 Nature of PD in rotating machines

4.1 Basics of PD
> ... where the dielectric properties of insulating materials are inhomogeneous ...
> ... local electrical over-stressing ...
> ... local, partial breakdown ...
> ... PD in general requires a gas volume to develop ...
> Gas filled voids embedded in the insulation,
> Voids adjacent to conductors
> At insulation interfaces
> ... numerous PD pulses during one cycle of the applied voltage ...
> ... Stator winding insulation ... for HV machines will normally have some PD .., but
are inherently resistant to partial discharges due to their inorganic mica
components...
> ... PD in these machines is usually more a symptom of insulation deficiencies ...
> .. PD ... may also directly attack the insulation and thus influence the ageing
process...
> ... The time to failure may not correlate with PD levels ...

4 Nature of PD in rotating machines

4.2 Types of PD
> ... For a given machine, the various PD sources may be identified and distinguished
in many cases by their characteristic PD behaviour...

> Internal discharges

>
>
>
>
>
>

Internal voids
Internal delamination
Delamination between conductors and insulation
Slot discharges
End-winding surface discharges
Conductive particles

4 Nature of PD in rotating machines

4 Nature of PD in rotating machines

A - Small voids on edge

B - Delamination: conductor main insulation
C - Delamination of tape layers
D - Treeing in layers
E - Broken strands
F - Micro voids
G - Slot discharge, semicon paint abrasion
H - Discharges in Cooling Duct
I - Delamination of Insulation in Elbow
J - End winding surface discharge - contamination
K - Insufficient Spacing, Tracking and Sparking
L - Connection area between slot corona protection and end winding corona protection

4 Nature of PD in rotating machines

4 Nature of PD in rotating machines

4.3 Pulse propagation in windings
> ... PD current ... is a transient pulse with a rise time of only a few
nanoseconds ...
> ... high frequency spectrum ...
> ... stator windings represent objects with distributed elements in which
travelling wave, complex capacitive and inductive coupling, and resonance
phenomena occur ...
> ... attenuation, distortion, reflection and cross-coupling ...
> ... transmission function from the PD source to ... sensor is unknown and
depends on the specific design of the machine ... Therefore, the energy at
the source of the PD, which can be taken as a measure of the erosion of the
insulation, cannot be measured directly ...

PD pulse propagation
Damping, reflection, attenuation

Amplitude of the pulse depends on:

>

Calibration

>

Ratio CK / CA

>

Center frequency

>

Propagation path
Two frequency components of the PD signals in a stator winding

4 Nature of PD in rotating machines

4.3 Pulse propagation in windings
> ... the individual high frequency
transmission behaviour of a stator
winding produces PD signals at the
terminals that are a characteristic
of the machine being tested and of
the location of the PD source ...

> ... very high frequency components

of PD signals are subject to
...attenuation ... and ... might not
be detectable at the terminals of
the test object ...

5 Measuring techniques and instruments

5.2 Influence of frequency response of measurement system
> ... measurement in the lower frequency range ensures good sensitivity
not only for partial discharges in bars/coils close to the sensor but also for
those that originate from further away in the winding. However, the lower
frequency range is more subjected to noise and disturbances ...
> ... measurement in the very high frequency range may acquire only a
very small proportion of the total PD energy, which results in sensitivity to
signals originating only very close to the sensor. However, this frequency
range may be less susceptible to noise and disturbance. ...
> ... For off-line PD testing to obtain appropriate sensitivity to PD from the
whole winding it is advisable to use wide band PD measuring systems. The
lower cut-off frequency should be in the range of several tens of kHz in
accordance with IEC 60270. ...

5 Measuring techniques and instruments

5.3 Effects of PD coupling units
> ... HV capacitor, coupling device, transmission system and input
impedance of the measurement system represent a high-pass filter.
Therefore, increased input impedance or higher capacitance values lead to
an increased sensitivity...

5 Measuring techniques and instruments

5.4 Wide-band and narrow band measuring systems

f >100kHz ... 1MHz

Superposition
errors!

6 Visualization of measurements
6.2 Data presentation
> PD magnitude [pC or mV]
> r.m.s. value of test voltage
> Inception voltage Ui (PDIV)

PDEV PDIV

> Extinction voltage Ue (PDEV)

> Curve Qm = f(U)

Specified threshold
for PDIV and PDEV
Noise level
U/Umax

6 Visualization of measurements
6.3 Additional means of PD data
representation
> ... Additional quantities ... like integrated
charge, discharge current, quadratic rate,
PD power, and PD energy ...
> ... pulse height distribution, pulse phase
distribution, phase resolved pulse height
distribution, oscillograms of pulse trains, PD
distribution maps, etc.... CIGRE technical
brochure 226 ...

Part V: Generator
Stator Insulation

6 Visualization of measurements
> 6.3.2. PRPD pattern
> ... a 2-dimensional PD distribution map (-q-n pattern) is employed for
visualization ...

7 Test circuits

IEC 60060-2:
Measuring systems

> HV power supply conforming to IEC 60060-1 and IEC 60060-2;

> A voltage measuring device;
> A suitable PD coupling unit;
> A connection cable from the measuring impedance to the PD device with
sufficiently low damping characteristics and good shielding;
> A partial-discharge measuring system;
> High-voltage connections.

> ... sufficiently PD free...

Example:
IEC 60060-1

7 Test circuits
> ... sufficiently PD free...
> ... To ensure that the test circuit does not influence the measurement of
partial discharges from the test object, the arrangement should first be
tested up to the maximum test voltage in accordance with the test
procedure given in 9.1.6. The noise level produced by the complete test
circuit at maximum required test voltage shall not exceed 100 pC when
using the normalization procedure in accordance with Clause 8...
> ... The whole test circuit should be of a low-inductance arrangement.
It is essential that ground loops are avoided. Low inductance leads are
recommended as ground connections...

7 Test circuits
Z

Ck
U~

Ca

CC
Zmi

CD

MI

7 Test circuits
Z

Ca

U~

Ck

CC
Zmi

CD

MI

7 Test circuits Complete winding

> ... HV source and the PD coupling unit should be connected to opposite
winding terminals whenever possible, to utilize the advantage of the
damping effect of the winding phases to suppress conducted interference
from the power supply...
> ... The PD coupling unit should be installed as close to the winding
terminals as possible...
> ... The stator core should normally be grounded...

7 Test circuits Complete winding

> Single winding energized, also measuring at the grounded windings

7 Test circuits Complete winding

Single winding
energized, also
measuring at the
grounded windings
(ground connection
at far end).

PD measurement on stator winding

Online measurement with permanent couplers

7 Test circuits Complete winding

Phases
Phasesnot
nottoto
be
beseparated
separated

Double length as
filter

8 Normalization of measurements
> ... Due to pulse propagation, resonance and mutual cross coupling in
machine windings, mentioned in 4.3, calibration is not possible...
> ... The aim of normalization is to ratio out various influences of the test
circuit, for example power supply connections, stray capacitance, coupling
capacitance and test object capacitance...
> ... Normalization is to ensure that the PD measuring system provides
sufficient sensitivity to measure a specified value of PD magnitude correctly,
as it appears at the machine terminals during the measurement, and to
show that the PD detection system used, is responding in a repeatable
fashion...
> ... comparisons between measurements on objects having the same
design, taken with the same PD device...
> ... Normalization of the test circuit should be performed by injecting shortduration current pulses of known magnitude by means of a reference pulse
generator (calibrator) ...

9 Test procedures
> Tests on complete windings
> Individual phases
> Individual winding components
> DUT disconnected from all external power supplies
>
>
>
>

Bus work
Surge arrestors
Surge capacitors
Excitation systems

> Test lead contact should always be at the machine terminals

> Same arrangement of all circuit components
> Same normalization procedure

9 Test procedures Equipment

> ... The waveform of the applied voltage should have
Upp/Urms = 2 SQR(2) , 5% ...

> ... it is acceptable to perform PD tests at lower frequencies ... or at higher

frequencies, 0.1Hz ... 400Hz

> ... PD results obtained from

very-low frequency tests
might significantly differ from
that at power frequency and
thus direct comparison might
not be possible ...

9 Test procedures Preparation

> ... stator should be inspected for cleanliness ...

> ... sufficient air clearance between ...connections and ... cables ...

> ... check the insulation resistance before starting the PD test ... >100M ...
[Mainly dirt, as dirt may influence PD behaviour]

> ... conductive slot coating should be in contact with ground potential
throughout its whole length ... [for winding components]

> CONDITIONING!
> ... PD will typically decrease during the first minutes ...
> ... 5 min at the maximum test voltage is recommended ...
> ... voltage may then be re-applied to start the partial discharge measurements
...

9 Test procedures Conditioning

Paschens Law

9 Test procedures Test voltages

> ... steps (e.g. U = 0,2 Umax) or ... continuous ramping (< 1 kV/s) up to ...
Umax. ...
> ... dwell time on each step of at least 10s is recommended ...

9.1.6 PD test procedure

> ... level of background noise...should be obtained to ensure that the test
arrangement has sufficiently low noise and PD up to the maximum test
voltage...
> ... replacing the test object by an appropriate discharge free capacitor ...
> ... Disturbances ... can be reduced or ... eliminated by appropriate
measures ...
> ... localize such disturbance signals ...
> ... temperature detectors (RTDs) are anchored to the grounded stator
frame ...
> ... the same connection point for the PD coupler, the test object and the
measuring equipment should be used ...
> ... test arrangement as compact as possible ...

[Link] PD testing
> ... each voltage step, or during continuous ramping, the PD data ... should
be recorded and processed ...
> ... Qm = f(U) curve ...
> PDIV
> PDEV
> ... distribution of pulse magnitudes, phase resolved distributions or specific
PD patterns ... at various levels during test voltage increase and decrease
...
> ... Any comments or observations during the test should be
recorded ...
> ... an appropriate sequence of standard measurements and ... extended
measurements ... should be taken ...

9.2 PD location and ID

> ... Following the per-phase test... energize simultaneously all three phases
of the winding ... [ No PD from end winding]
> ... subdivide the winding ...
> ... Electromagnetic probes ...
>
>
>
>
>
>

scanning of stator slots

machine end-winding
conduit boxes
cable routing
termination boards
stand-off insulators, etc.

> ... probes can disrupt the electric field ...

> ... additional safety requirements ...

10 Interpretation
> ... it has to be decided whether there are ... defects and ... what they imply
regarding the performance of the insulation system, whether any
supplementary tests are needed and the planning and/or implementation of
any essential corrective maintenance. ...

> ... significant variations in the amounts of partial discharges depending on

the individual properties of the machine being tested. As a result, a direct
comparison of different types of machine in terms of absolute values is not
possible. ...

> ... Neither is it possible to establish any absolute limits for complete
windings ...

10 Interpretation
> ... the PD site, which produces the highest PD magnitude is not necessarily
the location in the winding being at most risk ...

> ... PD magnitude as a function of test voltage provides a simple and

effective means of characterizing typical dominating PD sources ...

> ... Interpretation is always comparative ...

> ... meaningful interpretation for complete windings ...

> Trending Qm on the same stator over time
> Comparing Qm from several stators with the same design
> Comparing Qm between different phases of one stator

> ... the higher the PDIV and PDEV ... the less insulation deficiencies ...

10 Interpretation
Trend in PD in a machine over time
> ... Initial fingerprint ... when the winding is new ...
> ... If the winding deteriorates ... Qm will usually increase over time ...
> ... Compare the PD quantities ... between windings ...

> Limitations
> ... a new stator may have relatively high PD that decreases after the first 5000
to 10000 equivalent operating hours ...
> ... same voltage, temperature and similar humidity conditions ...
> ... variations of Qm of a certain percentage, for example 25 %, are normal
...

> Why?: ... That is, for example, Qm for the winding at the specified test
voltage is lower than 95 % of the mean Qm magnitudes achieved by the
manufacturer on the same windings they have made in the past....

10.3 Pattern recognition

> ... When using the -q-n patterns, it may be possible to separate various
PD sources from each other, to assess the related risk and to trend them
separately ...
> ... weight their risk separately ...
> Observe the trend
behaviour of each PD
source
> Localize the various PD
phenomena
> Provide rough
information concerning
location for pinpointing
> Assess the insulation
condition, depending on
PD source and PD
location.

3CFRD

Disturbance

PD

Noise

PRPD pattern shows multiple sources

L2

L1

L3

Separation of sources by 3PARD

3PARD = Three Phase Amplitude Relation Diagram

Inner PD source in L1 in 3PARD

Inner PD Source
L1>L2>L3
3PARD

L2

L1
Inner PD Source in
L1

L2

L3

L3

L1
timeframe
1 s

Outer noise in 3PARD

Outer Noise
L1 L2 L3

3PARD

L2

L1
Outer Noise

L2

L3

L3

L1
timeframe
1 s

PD measurement on Stator Winding and calibration

Localization: difficult
TDR

D L  Signal Velocity

TDD

 t1 
2

Signal Velocity

213

L  Signal Velocity t
 t unit1.1 

unit 1.2

c0
Hr Pr

Field diagnostics and diagnostic monitoring

of HV insulation at generators, motors,
transformers and cables (part II)

March 5, 2015

Topics
> Transformer
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities

> HV Cables
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities

Page
Page 22

Field Diagnostics & diagnostic monitoring of transformers

OMICRON

Seite 3

Life Time expectation

When a Transformer is operated under ANSI / IEEE basic loading conditions
(ANSI C57.96), its normal life expectancy is about 20 years. The ANSI / IEEE
basic loading conditions for Transformer are:
i. The Transformer is continuously loaded at rated kVA (kilo Volt Ampere)
and rated Voltages (Transformer must be operated at the rated Voltage
and kVA)
ii. The average temperature of the ambient air during any 24-hour period
is equal to 30C (86 F) and at no time exceeds 40C (104 F).
iii. The height where the transformer is installed, does not above 3300
feet or 1000 meters

OMICRON

Seite 4

Transformer Breakdowns

OMICRON

Seite 5

Accessories

Vessel & Oil

Core

Bushing

Winding

Tap
Changer

Faults in Transformers

Source: Cigre_WG 12-05 An international

survey on failures in large power
transformers in service, Electra No. 88, 1983

OMICRON

Seite 6

Insulation strength

Condition and Time based maintenance

New condition

Condition based maintenance

Diagnostic level
Time based maintenance
Operating stress
Time

Condition based maintenance

requires reliable diagnostic tools

Possible Impacts by lack of maintenance

Baseline data not recorded

Problems during warranty period not detected
OLTC contacts wear
Oil oxidation begins
Fan and Pump bearing wear
Visible effects of weathering and UV
Oil decay products affect paper insulation
Miss opportunity to intercept accelerated ageing
Oxidation and hydrolysis enters accelerated ageing state
OLTC and bushing failure rates increase
Dielectric withstand diminishes
Expensive failure (bushing, OLTC)

Life of the Transformer

OMICRON

Seite 8

Overview: Transformer Measurements

OMICRON

Seite 9

Maintenance Intervals
Action

Task Interval
Light

Regular

Remark
Intensive

Visit

6m

1m

1d

In service

Detailed inspection
(visual)

1y

3m

2w

In service

DGA

2y

1y

3m

Task interval may differ

with monitoring

Oil Test

6y

2y

1y

Cooling system
cleaning

Conditional

Conditional

Any interval

Outage may be required

Accessories
verification

12 y or
conditional

6-8 y

1-2 y

Outage required

Electrical basic tests

Conditional

Conditional

Any interval

Outage required

Insulation tests (DF)

Conditional

6-8 y

2-4 y

Outage required

OLTC inspection

12 y

6-8 y

4y

Overview: Transformer Measurements

Type of Problem
Magnetic Circuit Integrity
Magnetic Circuit Insulation
Winding Geometry
Winding/Bushing/OLTC Continuity
Winding/Bushing Insulation
Winding Turn to Turn Insulation

Diagnostic Technique
Basic Electrical

Winding Ratio

Winding Resistance
Magnetisation Current

x
x

Capacitance and DF/PF

x
x

Leakage Reactance
x

Advanced Electrical

Core Ground Test

Frequency Response of Stray Losses

x
x

x
x

Polarisation/Depolarisation

Frequency Domain Spectroscopy

Recovery Voltage Method

Electrical Detection of PD

Acoustical detection of PD

UHF Detection of PD

Dissolved Gas Analysis

OMICRON

Insulation Resistance

Frequency Response Analysis

Seite 11

Assessment and Interpretation

Indication

RBP

OIP

RIP

increase of capacitance

oil in cracks
or partial breakdowns

partial breakdowns

partial breakdowns

high dissipation factor

partial breakdowns;
insulator surface wet or
dirty (clean the insulator);
ageing of the inner
insulation;
water in the inner
insulation;

partial breakdowns;
insulator surface wet or
dirty (clean the insulator);
ageing of the inner
insulation;
water in the inner
insulation;

partial breakdowns;
insulator surface wet or dirty
(clean the insulator);
ageing of the inner
insulation;
water in the inner insulation;

dissipation factor is
decreasing with
increasing voltage

bad potential connections;

partial breakdowns

bad potential connections;

partial breakdowns

bad potential connections;

partial breakdowns

dissipation factor is
strongly increasing with
increasing temperature

high moisture in the

insulation;
high degree of ageing

high moisture in the

insulation;
high degree of ageing

high moisture in the

insulation;
high degree of ageing

normal, if constant

Discharges produce
gasses;
Errosion of the cellulose;
production of x-wax

partial breakdowns;
cracks or voids after
electrical or mechanical
stress;

partial discharges

Overview: Transformer Measurements

> Turns Ratio
> Exciting Current
> Winding Resistance (Dynamic and Static)
> Short Circuit Impedance (+FRSL)
> C & DF (Winding, Bushing and Oil sample)
> Moisture Determination in liquid and solid insulation
> Frequency Response Analysis Partial Discharge

OMICRON

Seite 13

Power transformer diagnosis

Transformer power and dissipation factor (tan delta) + insulation capacitance
Automated testing with test templates and reporting with ExcelTM

Capacitance
(voltage sweep)
30nF

0.45%

25nF

V
H(V)

OMICRON

V
HL(V)

L(V)

H(V)

HL(V)

L(V)

14000V

12000V

10000V

14000V

12000V

10000V

8000V

0nF
6000V

0.2%
4000V

5nF
2000V

0.25%

8000V

10nF

6000V

0.3%

15nF

4000V

0.35%

20nF

2000V

0.4%

0V

Capacitance

0.5%

0V

Power factor / dissipation

factor (tan delta)

Tan delta
(voltage sweep)

Advanced diagnostics
Transformer power and dissipation factor (tan delta)
with variable frequency from 15 Hz to 400 Hz

Transformer with aged oil

New transformer

f
H(f)

HL(f)

L(f)

H(f)

Additional diagnostics revealed through

frequency sweep from 15 Hz to 400 Hz.

OMICRON

HL(f)

L(f)

450Hz

400Hz

350Hz

0%
300Hz

450Hz

400Hz

350Hz

300Hz

250Hz

200Hz

150Hz

100Hz

50Hz

0Hz

0%

0.2%
0.1%
250Hz

0.1%

200Hz

0.2%

0.4%
0.3%

150Hz

0.3%

100Hz

0.4%

0.6%
0.5%

50Hz

0.5%

Increased power factor /

dissipation factor especially
at low frequencies

0.7%

0Hz

Power factor / dissipation

factor (tan delta)

Power factor / dissipation

factor (tan delta)

0.6%

Advanced diagnostics
Transformer power and dissipation factor (tan delta)
with variable frequency from 15 Hz to 400 Hz
Advantages

> Faster and more reliable assessment of

transformer and bushing conditions (aging)
> Shows additional details not obtainable
from single power factor measurements:
> Determines whether moisture contamination is in the cellulose
or if the oil is contaminated or otherwise compromised.
> Effective discrimination between a deteriorated power factor test result
that warrants more frequent monitoring and one that requires immediate
remediation.
> Faster fault localization

OMICRON

Power transformer diagnosis

Bushing power and dissipation factor (tan delta) + insulation capacitance

Bad bushing

0.32%

Power factor / dissipation

factor (tan delta)

0.31%
0.3%
0.29%
0.28%
0.27%
0.26%
0.25%

Inmost layer was not

connected to the HV
conductor properly.

1.1%
1%
0.9%
0.8%
0.7%
0.6%

OMICRON

14000V

12000V

10000V

8000V

6000V

4000V

14000 V

12000 V

10000 V

8000 V

6000 V

4000 V

2000 V

0V

2000V

0.5%

0.24%

0V

Power factor / dissipation

factor (tan delta)

New bushing

Advanced diagnostics
Bushing power and dissipation factor (tan delta)
with variable frequency from 15 Hz to 400 Hz

Condenser type bushing (220 kV RIP)

Power factor / dissipation
factor (tan delta)

0.85%
0.75%
only fNOM values

0.65%

Aged bushing

0.55%

Medium-aged bushing

0.45%

New bushing
0.35%
0.25%
0Hz

100Hz

200Hz

300Hz

400Hz

500Hz

Additional diagnostics revealed through

frequency sweep from 15 Hz to 400 Hz.

OMICRON

Advanced diagnostics
Bushing power and dissipation factor (tan delta)
with variable frequency from 15 Hz to 400 Hz

Oil-impregnated bushings (OIP)

Power factor / dissipation
factor (tan delta)

0.6 %

0.55 %

0.5 %

0.45 %

450 Hz

400 Hz

350 Hz

300 Hz

250 Hz

200 Hz

150 Hz

100 Hz

50 Hz

0 Hz

0.4 %

f
1

Looking at the whole frequency range shows that four bushings

have aged considerably. At lower frequencies there is an
extreme increase in the dissipation factor (tan delta).

OMICRON

Power transformer diagnosis

Insulating fluids (oil test cell) power and dissipation factor (tan delta)

Testing the insulation condition of oil

OMICRON

Power transformer diagnosis

Exciting current per tap

Exciting current per tap

changer position

Exciting current and phase

angle per tap

0.0045 A

Exciting current

0.004 A
0.0035 A
0.003 A
0.0025 A
0.002 A
0.0015 A
0.001 A
0

10

15

20

25

Tap changer position

A

OMICRON

30

Power transformer diagnosis

Exciting current per tap

Exciting current

100 mA
80 mA
60 mA

40 mA

B
C

20 mA
0 mA
1

10

11

Tap changer position

1000 W

Watt losses

800 W
600 W

400 W

B
C

200 W
0W
1

Tap changer position

OMICRON

10

11

Measurement and report table

OLTC
DETC
Position Position

Phase A
I out
[mA]

Watt
Loss
[W]

Reactance Phase B
(+/-jX)
I out
[k]
[mA]

Watt
Loss
[W]

Reactance Phase C
(+/-jX)
I out
[k]
[mA]

Watt
Loss
[W]

Reactance
(+/-jX)
Pattern
[k]
Rating

61,56

573,22

67,62

42,81

396,23

88,42

59,00

540,12

67,43

NONE

62,62

569,87

66,09

44,31

410,38

86,00

61,12

559,15

66,15

NONE

64,79

588,69

64,45

45,79

422,97

83,65

63,23

577,36

64,53

NONE

67,05

608,41

62,72

47,44

438,60

81,21

65,46

596,85

62,81

NONE

70,51

625,79

60,10

49,08

452,48

78,84

67,80

617,27

61,04

NONE

71,89

651,07

59,04

50,92

469,99

76,44

70,24

638,84

59,23

NONE

74,44

673,86

57,11

52,74

486,24

73,79

72,79

661,55

57,33

NONE

77,18

698,20

55,26

54,69

503,53

71,31

75,56

685,89

55,56

NONE

80,07

724,07

53,37

56,78

522,56

68,86

78,47

711,65

53,72

NONE

10

82,97

750,83

51,35

58,97

542,90

66,22

81,55

739,10

51,88

NONE

11

86,33

780,10

49,64

61,28

564,43

63,40

84,81

768,08

50,05

NONE

Power transformer diagnosis

Transformer turns ratio (TTR) per tap

Settings: transformer ratio per tap

17

Ratio

16
15

A
B

14

13
12
000

005

010
Tap

OMICRON

015

020

Power transformer diagnosis

Transformer turns ratio (TTR) per tap

Transformer in good condition

Ratio deviation

0.04 %

TRRatio:

0.02 %
0.00 %

-0.02 %

-0.04 %
C

-0.06 %
-0.08 %
0

10

15

20

Taps

Exciting current

0.01 A

NO

Assessment:
Range:

0.004 A
0.002 A
0

10
Taps

15

20

n/a
AC 2kV

Nominal
values:
Frequency:

50,0Hz

V test:

1000,0V

Tap

V prim. Nom.

Vsec.
nom.

Ratio
nom.

V prim.

001

449610,0/3V

15750,0V

16,4814:1

999,59V

0,0

60,636585V

0,0

16,4849:1

-0,02%

0,004566A

-40,71

002

444890,0/3V

15750,0V

16,3084:1

999,74V

0,0

61,285938V

0,01

16,3127:1

-0,03%

0,004625A

-41,18

003

440170,0/3V

15750,0V

16,1354:1

999,79V

0,0

61,947819V

0,0

16,1392:1

-0,02%

0,004721A

-41,48

004

435440,0/3V

15750,0V

15,962:1

999,85V

0,0

62,622185V

0,0

15,9664:1

-0,03%

0,004791A

-41,86

005

430720,0/3V

15750,0V

15,789:1

999,83V

0,0

63,309776V

-0,01

15,7927:1

-0,02%

0,004902A

006

426000,0/3V

15750,0V

15,616:1

999,81V

0,0

64,010597V

-0,01

15,6194:1

-0,02%

0,005008A

-42,35

007

421280,0/3V

15750,0V

15,4429:1

999,78V

0,0

64,727524V

-0,01

15,446:1

-0,02%

0,005097A

-42,77

008

416560,0/3V

15750,0V

15,2699:1

999,89V

0,0

65,464005V

-0,01

15,2739:1

-0,03%

0,005211A

-42,96

009

411830,0/3V

15750,0V

15,0965:1

999,6V

0,0

66,197601V

-0,01

15,1002:1

-0,02%

0,005334A

-42,71

010

407110,0/3V

15750,0V

14,9235:1

999,78V

0,0

66,981621V

0,01

14,9262:1

-0,02%

0,005422A

-43,33

0,00568A

-45,03

15750,0V

14,5775:1

999,83V

0,0

68,570297V

0,0

14,5811:1

-0,02%

0,005816A

-45,37

15750,0V

14,4044:1

999,78V

0,0

69,396141V

-0,02

14,4069:1

-0,02%

0,005939A

-45,7

014

388230,0/3V

15750,0V

14,2314:1

999,74V

0,0

70,235367V

0,0

14,2341:1

-0,02%

0,006073A

-45,83

0,0

14,0614:1

-0,02%

0,006229A

-46,1

016

378780,0/3V

15750,0V

13,885:1

999,72V

0,0

71,986328V

0,0

13,8876:1

-0,02%

0,006361A

-46,33

017

374060,0/3V

15750,0V

13,712:1

999,78V

0,0

72,902046V

-0,01

13,714:1

-0,01%

0,00651A

-46,59

018

369340,0/3V

15750,0V

13,539:1

999,79V

0,0

73,832764V

-0,01

13,5413:1

-0,02%

0,006659A

-46,73

019

364620,0/3V

15750,0V

13,3659:1

999,75V

0,0

74,779701V

-0,01

13,3693:1

-0,02%

0,006823A

-46,88

14,058:1

999,73V

999,79V

0,0

0,0

67,760643V

71,101517V

0,0

14,7538:1

-0,02%

-42,1

397670,0/3V

15750,0V

14,7505:1

I prim.

392950,0/3V

383500,0/3V

15750,0V

Ratio

012

015

402390,0/3V

Vsec.

013

0.006 A

OMICRON

02.05.2008 13:22

Overload:

011

0.008 A

Date/Time:

Power transformer diagnosis

Transformer turns ratio (TTR) per tap

Automated measurement with switch box CP SB1

> Tremendous time-saver: removes the need

to change test connections between tests
> Automatically operates the tap changer
> Quickly discharges the transformer
> Increases safety by reducing the risk of accidents
> Minimizes the likelihood of measurement errors
> Automated ratio and winding resistance tests
> Manual operation mode: allows measurements
for special applications (e.g. auto transformer tests)

OMICRON

Power transformer diagnosis

Leakage reactance / short circuit impedance

Determination of winding or core deformation due to high fault currents

OMICRON

Power transformer diagnosis

Leakage reactance / short circuit impedance

Equivalent circuit diagram:

effective resistance and reactance

Leakage flux

Forces

10 kV winding

220 kV winding

Forces in a transformer, especially in case of a

short circuit, can damage windings and the core.

OMICRON

Power transformer diagnosis

Leakage reactance / short circuit impedance

Factory Test with W-Meter

Factory test data

3-Phase Measurement

SN [kVA]

Pk [kW]

f [Hz]

Tap Position:

UHV [kV]

ULV [kV]

uk [%]

Uk [kV]

6500

34.5

50

51.7

27.2

6.250

1.9

Three Phase Equivalent Test

CPC 100 Measurement

f [Hz]

V1 AC [V]

I AC [A]

Z []

Phi []

R []

X []

L [mH]

A-B

50

55.224

1.049

25.7019

85.01

2.2370

25.6044

81.50

B-C

50

54.675

1.048

25.8922

85.24

2.1470

25.8030

82.13

A-C

50

54.464

1.048

25.2547

84.96

2.2196

25.1570

80.08

Average Impedance
f [Hz]

uk [%]

Z []

Phi []

R []

X []

xk [%]

L [mH]

Factory Test Values:

50

6.250

25.7009

85.13

2.1826

25.6080

6.227

81.51

Measured Values:

50

6.229

25.6162

85.07

2.2012

25.5215

6.206

81.24

-0.33

-0.33

-0.07

0.85

-0.34

-0.34

-0.34

Difference to Factory Test [%]

Per-Phase Test

Factory test and

measurement comparison

CPC 100 Measurement

f [Hz]

V1 AC [V]

I AC [A]

Z []

Phi []

R []

X []

L [mH]

A-B

50

25.061

1.010

24.5892

84.37

2.4102

24.4708

77.89

B-C

50

25.337

1.011

24.8397

84.54

2.3641

24.7270

78.71

A-C

50

25.485

1.010

24.9888

84.44

2.4224

24.8711

79.17

24.8059

2.3989

24.6896

78.5896

0.87

1.45

0.89

0.89

Average Impedance

Maximum delta of Average [%]

OMICRON

Measurement in accordance
to IEEE Std C57.12.90

Per-phase test

Advanced diagnostics
Frequency response of stray losses (FRSL)

Measured resistance

This test is automatically performed with the

leakage reactance / short circuit impedance test.

Frequency

R is frequency-dependent due to eddy losses (induced currents)

in the windings which will increase with the frequency.

OMICRON

Advanced diagnostics
Frequency response of stray losses (FRSL)

Parallel strands without twisting

HV winding
LV winding
Losses through
induced currents

OMICRON

Advanced diagnostics
Frequency response of stray losses (FRSL)

B
HV winding

LV winding
Parallel strands
with twisting

Induced currents
are compensated

Parallel strands with twisting compensate

the induced currents.

OMICRON

Advanced diagnostics
Frequency response of stray losses (FRSL)

B
HV winding

LV winding
Additional losses through
induced currents

If the twisted strands are shorted additional

losses will occur due to induced currents.

OMICRON

Advanced diagnostics
Frequency response of stray losses (FRSL)

Three phase equivalent

Three phase equivalent

5
20.4mH
4
20mH

3

19.6mH

2

19.2mH

1

18.8mH

0
0Hz

100Hz

200Hz

300Hz

400Hz

0Hz

100Hz

200Hz

ZkBC

ZkAC

ZkAB

ZkBC

The deviation of the resistance Rk,AC indicates short

circuit parallel strands in the left diagram.

OMICRON

400Hz

f
ZkAB

300Hz

ZkAC

Power transformer diagnosis

DC winding resistance

Measurement of winding resistance and internal contacts

Tap selector

Tap changer

OMICRON

Connection clamps

Power transformer diagnosis

DC winding resistance

Measurement example
Tap changer in
good condition

Defective contacts in the

tap selector
0.19
Winding resistance

Winding resistance

0.19
0.18
0.17
0.16
0.15

0.18
0.17
0.16
0.15
0.14

0.14
0

10

15

20

25

30

10

Taps

20

Taps

Rmeas

The contacts which were used in taps

1, 3, 5, 23, 25, and 27 were defective.

OMICRON

15

Rref

Rmeas

25

30

Power transformer diagnosis

Transformer demagnetization

Influence of remanence

Exciting current
0.02A
0.018A
0.016A
0.014A

Frequency response anaylsis

U with remanence
V with remanence
W with remanence
U without remanence
V without remanence
W without remanence

0.012A
0.01A
0.008A
0.006A
0.004A
0.002A
0.0A

OMICRON

Initial FRA verification

After DC test
After demagnetization

Advanced diagnostics
Dynamic LTC diagnostics with CP SB1 (OLTC test)

OMICRON

Advanced diagnostics
Dynamic LTC diagnostics with CP SB1 (OLTC test)

Transient current during

switching process
5.1 A
Slope

Diverter switch switches to the

first commutating or transition resistor

Current

5.0 A

Final contact of the diverter contact B is reached

4.9 A
Ripple
4.8 A

4.7 A
-0.05 s

0s

0.05 s 0.1 s 0.15 s 0.2 s 0.25 s 0.3 s 0.35 s

Time

Both commutating resistors are in parallel

Current control of the CPC 100 regulates

the current to the nominal test current again

The transient switch is recorded and graphically displayed

to find faults in the OLTC, tap selector, and DETC.

OMICRON

Advanced diagnostics
Dynamic LTC diagnostics with CP SB1 (OLTC test)

Aged diverter switch

2.5%

5.5%

2%

5%
Ripple

Ripple

Diverter switch in good

condition

1.5%
1%
0.5%

4.5%
4%
3.5%

0%

3%
0

10

15

Taps

15

20

Taps

AUP

ADOWN

BUP

AUP

ADOWN

BUP

BDOWN

CUP

CDOWN

BDOWN

CUP

CDOWN

Ripple and slope show similar results.

OMICRON

10

Field Diagnostics & diagnostic monitoring of HV Cables

OMICRON

Seite 40

Topics
> Transformer
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities

> HV Cables
Lifetime expectations
Faults and Aging
Measurement & monitoring parameters /quantities

Page
Page 41
41

Life Time expectation

Similar to other HV components e.g. power transformers, circuit

breakers and overhead lines the average service age of transmission
power cable networks is between 30-45 years. Moreover, no or limited
knowledge exists about such future performances as:
- insulation degradation processes,
- operation reliability,
- maintenance /replacement expectations.

Cigre WG 37 -27, Aging of the System - Impact on Planning, Cigre Brochure 2001

OMICRON

Seite 42

Faults and aging

Repair costs of a failure in a HV cable link can be estimated of

around 500 /kV. This means that the repair of a failure in a 150kV
link costs around 75 k. In addition also costs of not delivered energy
have to be taken in to account and these additional costs, depending
on claims and penalties as contracted can be much higher.

Popma J. and Pellis J., Diagnostics for high voltage cable systems, proceedings ERA
conference on HV plant life extension, Belgium, 23-24 November, 2000

OMICRON

Seite 43

Diagnostic Parameters
Contribution of different on-site diagnostics to short and/or long term
condition assessment
a) measurement of dielectric
losses/response
b) PD measurements
c) Dissolved gases analysis
d) DC sheath test
e) visual inspection of accessories
f) Inspection of earthing system
g) Inspection of hydraulic system
h) determination of impregnation
coefficient

Popma J. and Pellis J., Diagnostics for high voltage cable systems, proceedings ERA conference on
HV plant life extension, Belgium, 23-24 November, 2000

OMICRON

Seite 44

Diagnostic Parameters

E. Gulski, B. Quakers Groot, Th Strehl, E. Lemke, et al, Fundamental Aspects of Data

Quality for HV Asset Condition Assessment, Cigre ELECTRA No 228 October 2006

OMICRON

Seite 45

Technologies for diagnostic monitoring of

HV insulation at generators, motors,
transformers and cables

March 5, 2015

Topics
> Diagnostic Monitoring Systems
> Permanent and Temporary PD Monitoring System for Rotating
Machines
> PD, C, TanDelta, transient Monitoring System for Transformers
> Permanent and Temporary PD Monitoring System for HV Cables

Page
Page 22

Measurement vs. Monitoring

> Monitoring enables measurement comparisons to be made in the
presence of all operational stress factors (3-phase, asymmetries,
couplings, loads, vibrations, temperatures, etc.) over longer periods of
time.
> Monitoring helps to predict severe failure to extend reliable operation of
equipment throughout its service life.

OMICRON

Page 3

Diagnostic monitoring
What is diagnostic monitoring?
> The large amount of real-time (trend) data gathered by monitoring
systems can be combined with precise insulation condition assessment
> Effective comparison of historical data is enabled via an easy-to-use and
extendable database solution
Benefits for asset management
> Monitoring answers questions about the present condition of the
equipment and its future performance
> Exact knowledge of insulation state at any time saves money, as
maintenance schedules can be specifically optimized and the service life
of assets can be extended
> Periodic or continuous monitoring of HV equipment is an additional,
essential tool for effective maintenance management

OMICRON

Page 4

Diagnostic monitoring
> Involves no operational variables
> Provides information to evaluate the aging condition of equipment
> Allows assessment of future developments in the aging process
> Supports decision making for ongoing maintenance or the replacement
of components
> is a "planning tool (asset management)

OMICRON

Page 5

Diagnostic monitoring
> Continuous or periodic monitoring of HV equipment is an essential tool for
effective maintenance management
> Monitoring answers questions about the present condition of the
equipment and its future performance
> The large amount of real-time data gathered by monitoring systems can
be used for precise insulation condition assessment
> Effective comparison of historical data is enabled via an easy-to-use and
extendable database solution
> Exact knowledge of insulation state at any time saves money, as
maintenance schedules can be specifically optimized and the service life
of assets can be extended

OMICRON

Page 6

Diagnostic monitoring
> Involves no operational variables
> Provides information to evaluate the aging condition of equipment
> Allows assessment of future developments in the aging process
> Supports decision making for ongoing maintenance or the replacement
of components
> is a "planning tool (asset management)

OMICRON

Page 7

Partial discharge (PD) monitoring

> Partial discharge (PD) phenomena are a consequence of local electrical
stress concentration in electrical insulation.
> PD activity can lead to failure and serious consequent damage as well as
to faults in critical elements of the power network.
> Therefore it is mandatory to identify the presence of PD activities and to
monitor PD tendencies.
> PD is a widely-accepted measuring parameter for insulation diagnosis.
> PD measurements are specified for type, routine and on-site tests for
most HV assets.
> The large variety of PD signals makes PD measurement and detection
a challenging task.
> OMICRON monitoring systems employ advanced noise suppression
techniques for reliable PD detection.

OMICRON

Page 8

Requirements of a monitoring system

> Long lifetime of the sensor system (comparable to that of the
monitored asset)
> Minimal physical and temporal interference in existing installations
> Does not influence other online or offline measurements / monitoring
processes (e.g. suitability for offline tests with elevated voltages)
> Does not endanger personnel, equipment and infrastructure
> No potential risk to internal IT structures
> Low operating costs; no maintenance required
> Flexible alarm, monitoring and information system
> Possibility for remote access by resource experts
> Reproducibility and comparability of measurement results

OMICRON

Page 9

Topics
> Diagnostic Monitoring Systems
> Permanent and Temporary PD Monitoring System for Rotating
Machines
> PD, C, TanDelta, transient Monitoring System for Transformers
> Permanent and Temporary PD Monitoring System for HV Cables

Page
Page 10
10

PD Monitoring - Rotating machines

Copyright "Siemens Pressebild

OMICRON

Seite 11

On-line monitoring on rotating machines

WHERE?
Stator Windings
WHAT?
Partial Discharge

OMICRON

Seite 12

OMS 600
Permanent PD monitoring for rotating machines

OMICRON

10 March 2015

Permanent monitoring solution: OMS 600

Benefits
> Continuous assessment of insulation state
in stator windings
> Early defect detection prevents failures
> Synchronous multi-channel PD data acquisition
> Advanced technology like 3PARD and 3CFRD
for PD source separation
> De-noising using several gating methods
(amplitude-, phase-, dynamic-, frequency-,
channel-gating)
> Optional temperature monitoring can be
added for correlation with PD data
> Web-based data access, management
and visualization

OMICRON

Page 14

OMS 600 continuous monitoring system

for motors and generators

Permanent monitoring solution: OMS 600

Extendable system for one or several machines at one location
1st machine

1st machine

Capacitive
sensors
Acquisition
unit

Fiber optic
cable

Central computer
and software
OMICRON

Page 15

2nd machine

Additional
machines

Permanent monitoring solution: OMS 600

1

Capacitive sensors

Fiber optic bus

Data acquisition unit

Central computer and software

Fiber optic communication

5
1

1
2
3

OMICRON

Seite 16

OMS 600: System components

MCC 112 / 117 / 124 capacitive sensors
> 3 capacitive sensors One for each phase of the machine
> Installation within the generator's main terminal box
to decouple the PD signal
> Easy to integrate into new / already in-service machines
> Connected to the data acquisition unit via screened triaxial
cables in order to minimize the impact of external interference

Kapazitiver Sensor
OMICRON

Seite 17

OMS 600: System components

OMS 600 data acquisition unit
> Synchronous multi-channel
PD data acquisition
> Advanced technology like 3PARD and
3CFRD for PD source separation
> Optical isolation between partial
discharge acquisition unit and server
using fiber optic cable
OMS 600
Data acquisition unit

OMICRON

Seite 18

OMS 600: System components

Central computer and monitoring software
> State-of-the art database system for
long-term data storage and retrieval
> Web-based data access & visualization
> Different user roles/logins:
> Operators:
> Overview of full system setup
> View all data with free configurable charts
and diagrams
> Confirm warnings and alarms

> Administrator: As operator, plus

> Configuration of all monitoring parameters
> Set and change warning and alarm rules

OMICRON

Seite 19

Online monitoring on rotating machines

Challenges of PD Monitoring?
> Partial Discharge (PD) is tolerable phenomena
(PD resistive isolation)
> BUT depending on location and characteristics
they might be the cause or a symptom of
isolation failure
> Differentiation between potentially dangerous
and harmless PD
> Data Interpretation
> various external noise sources
> many active PD sources in parallel
> overlapping of sources in PRPD

OMICRON

Seite 20

Online monitoring on rotating machines

Challenges of PD Monitoring - Frequency Spectrum Detection Ranges

UHF

OMICRON

Seite 21

Online monitoring on rotating machines

Challenges of PD Monitoring Winding Coverage

350

100

MHz

40

20

0.5

0.025

10

15

20

25

30

35

40

45

50

% Relative Coverage

OMICRON

Seite 22

55

60

65

70

75

80

85

90

95

100

Monitoring systems by OMICRON

PD monitoring for rotating machines
> On-line condition assessment of stator
insulation during standard operating conditions
> Very sensitive measuring technology, e.g.
> High capacitance of coupling capacitors
> Adjustable frequency bandwith
(=> best Signal to Noise Ratio)
> Synchronous 3-phase PD Measurement

> Early detection of PD activities and trend

display to prevent machine failure
> Powerful analysis software and server
architecturefor long-term data storage,
visualization & evaluation

OMICRON

Seite 23

Monitoring systems by OMICRON

PD monitoring for rotating machines
> 3PARD (3-Phase Amplitude Relation Diagram)
> state-of-the-art diagnostical tool
> display of the relation of PD pulses detected in
all
3 phases (time synchronous measurement)
> separation of PD sources and related patterns
(1 cluster per source)
> Elimination of overlapping effects

PD sources with highest amplitudes are not

always the most harmful ones!
> Support of condition-based maintenance to
extend the service life of generators and
motors
> Customized & expandable solution

OMICRON

Seite 24

Partial discharge in stator windings

Principle of stator winding insulation
A) Covering Tape
B) Spacer, Coil-End Bracin
C) Groundwall (main) Insulation, Mica Tapes
D) Grading/ Silicon Carbide Coating
E) Slot Semiconductive Coating
C

F) Inner Semicon. coating

G) Turn Insulation

H) Slot Wedge / Seal

I) Stator Core

OMICRON

C
C

Seite 25

Partial discharge in stator windings

Cause of PD in stator windings
> Damaging of Insulation through combined forces
Highest and lowest temperature
Increase of Current through Overload
Rated Voltage
Test Voltage
Over Voltage
water, ice
aggressive and reactive chemicals: gas, acids
abrasive particles: metal parts, ash, carbon
Oscillation in slot section
mechanical forces in enwinding section
Different thermal expansion

OMICRON

Seite 26

Partial discharge in stator windings

Cause of PD in stator windings
A: small voids on edge
B: Delamination: conductor main insulation
C: Delamination of tape layers
D: Treeing in layers
E: Broken strands
G: Micro voids
F: Slot discharge, semicon paint abrasion
H: Discharges in Cooling Duct
I: Delamination of Insulation in Elbow
J: Endwinding surface discharge - contamination
K: Insufficient Spacing, Tracking and Sparking
L: Connection area between slot corona protection and endwinding
corona protection

OMICRON

Seite 27

high

Delamination: conductor
main insulation

normal

J&L

Surface discharges end winding

Micro void, different scaling

corona

normal

Micro void, different y- Axis scaling

medium

Inner PD + Endwinding surface

discharge - contamination

low

high

Inner PD + Endwinding surface

discharge - contamination

Slot discharge, semicon paint

abrasion

Insufficient Spacing, Tracking and

Sparking,
winding
to
core
clamping bolts or between two
phase windings

Corona, winding to core

clamping bolts

low

low

normal

Micro void

noise

noise-excitation (also 12 and 24

peaks)

high

noise

Noise, asynchronous with AC

low

Inner PD + Endwinding surface

discharge - contamination

Delamination of tape layersbigger inner micro void

low

Inner PD + Endwinding surface

discharge - contamination

Contact PDs

Challenge: Overlapping of PD sources

Monitoring Software Display
Noise

200 pC

Noise, asynchronous with AC

medium

900 pC

Inner PD + Endwinding surface

discharge - contamination

Alarm

high

500 pC
Delamination of tape layersbigger inner micro void
OMICRON

Time
Seite 29

Challenge: Overlapping of PD sources

3PARD Analysis Principle
Capacitive
Capacitive
&&Inductive
Inductivecoupling
couplingtotothe
thewindings
wingdings
Capacitive
coupling
Inductive
Capacitivecoupling
& Inductive coupling
ofofthe
same
other
phase
two phases

> 2015-03-10

E
E

36

35

34

33

32

31

30

TE

CORE

L1
L2
L3
neutral

29

28

27

26

25

24

23

22

21

20

19

18

17

16

15

14

13

12

11

10

36

35

34

33

32

Challenge: Overlapping of PD sources

3PARD Analysis Principle

Inner PD-source
L1>L2>L3
L1

L2
Inner PD source in L1

L2

L3

L3

L1
timeframe
1 s

Because of the complex propagation and coupling characteristics of PD impulses, every source
of PD impulses has unique impulse to couplings ratio and consequently a characteristic 3PARD
signature
Seite 31
OMICRON

3/10/2015

Challenge: Overlapping of PD sources

3PARD Analysis Principle

Noise
L1 L2 L3
L1

L2
noise

L2

L3

L3

L1
timeframe
1 s

Electrical interferences and outer noise usually have the pulse-like nature of PD but
occur with almost same amplitude on all three phases of the machine.

OMICRON

Seite 32

Challenge: Overlapping of PD sources

3PARD Analysis Principle

OMICRON

Seite 33

Temporary and permanent monitoring solutions

Temporary monitoring system

Permanent monitoring system

OMS 605
Temporary PD monitoring for rotating machines

OMICRON

10 March 2015

Temporary monitoring solution: OMS 605

Why periodic PD monitoring with OMS 605?
> Partial discharges (PD) indicate damaged or aging insulation
> Detects insulation defects early to prevent asset failure
> Indicates which HV assets need to be monitored on a continuous basis
> Supports cost effective, condition-based maintenance to extend asset life

OMICRON

Page 36

Temporary monitoring solution: OMS 605

Benefits
> 3 channels for synchronized
PD data acquisition
> Plus one channel for additional
acquisition unit
> Wheeled case enables easy transport,
to different monitoring locations
> Rugged measurement system
for industrial environments (IP65)

OMS 605 portable PD monitoring system

for high-voltage assets

> Simple to use plug-and-play modules

for fast system set-up
> Compatible to a variety of sensors
> Powerful software for PD analysis,
trending and data visualization
OMICRON

Seite 37

Temporary monitoring solution: OMS 605

Portable system for one or more machines

Coupling
capacitors

Terminal box

OMS 605

Notebook
with software

OMICRON

Seite 38

OMS 605: System components

1

Fiber optics cable

Signal cable

Grounding band

USB cable

Grounding clamp

Power supply cord

2
1
4
3

5
6

OMICRON

Seite 39

OMS 605: System components

1

Top cover tray

CAL 542 calibrator

Mobile workstation (optional)

MCU 502 controller

MCU external socket

3
4

OMICRON

Seite 40

OMS 605: System components

MCC 124 / MCC 117 / 112 capacitive sensor set
> 3 capacitive sensors are mounted within the
generator's main terminal box to decouple PD
signal
> Easily integrateable into new / already in
service machine
> The capacitive sensors are connected to
the OMS 605 system via screened triaxial
cables in order to minimize the impact of
external interferences

OMICRON

Page 41

OMS 605: System components

Sensor transportation cases for multiple-asset monitoring (optional)
4 x MCC 117 + 4 x MCT XXX

OMICRON

3 x MCC 124 + 4 x MCT XXX

Seite 42

OMS 605: System components

Terminal box for rotating machines
> Convenient sensor connections for
temporary PD monitoring with the
OMS 605
> Plug-and-play connection to
permanently-installed coupling
capacitors

3-channel

> Available with either 3-channel

inputs/outputs or 4-channel
inputs/outputs
> IP 66 (EN 60529)

OMICRON

Seite 43

4-channel

OMS 605: System components

Mobile monitoring software
1

Set up a monitoring session

> Select an existing asset template

or create a new one
> Define monitoring duration and
frequency
> Set warning and alarm thresholds
> Import settings from previous
monitoring sessions or diagnostic
tests

OMICRON

Seite 44

OMS 605: System components

Mobile monitoring software
1

Start a monitoring session

> View running monitoring session

> Capture pre-analyzed PD trend
data
> Record 30-second data streams
to be analyzed using the expert
PD software

OMICRON

Seite 45

OMS 605: System components

Mobile monitoring software
2

View PD trend data from

completed sessions

> Warning / alarm indication notifies

you when user-defined thresholds
have been exceeded
> Scroll over data points to view
time and value of PD activity
> Click on data points to display
PRPD and 3PARD images
> Trend data can be easily exported
in a variety of formats for reporting

OMICRON

Seite 46

OMS 605: System components

Mobile monitoring software
2

View PD trend data from

completed sessions

> PRPD (Phase-resolved PD)

and 3PARD (3-Phase
Ampliture Relation Diagram)
images show overlapping
and separated PD activity

PRPD

> Trend data can be easily

exported in a variety of
formats for reporting

3PARD
OMICRON

Seite 47

OMS 605: System components

Expert PD analysis software
3

Real-time diagnosis of PD data

> Various filters can be applied to

see more detail of PD activity
> Resulting information helps you to
reliably assess insulation
condition status
> Data can be exported for reporting
in a variety of file formats (i.e.
Excel, MATLAB)

OMICRON

Seite 48

Topics
> Diagnostic Monitoring Systems
> Permanent and Temporary PD Monitoring System for Rotating
Machines
> PD, C, TanDelta, transient Monitoring System for Transformers
> Permanent and Temporary PD Monitoring System for HV Cables

Page
Page 49
49

Field Diagnostics & diagnostic monitoring of transformers

OMICRON

Seite 50

On-line monitoring on power transformers

WHERE?
> Bushings
> Transformer tank
WHAT?
> Capacitance
> Dissipation/power pactor
> Partial discharge (PD)
> Transient over-voltages

OMICRON

Seite 51

Theory
Supplementary bushing monitoring is vital
> Bushing failures contribute significantly
to failure statistics of power transformers
> With dissolved gas analysis (DGA), only
by-products of insulation degradation
inside the tank can be detected not at
the bushings

Cigre A2.37 Transformer

Failure Statistics

Tap changer
26%

Damaged transformer after

a bushing explosion

Other
1%

Failure Locations of Substation

Transformers >100 kV
Core and
magnetic
circuit
3%

Bushings
17%

Source: No. 261 April 2012 ELECTRA

Isolation
1%

OMICRON

Page 52

Lead exit
7%

Winding
45%

Theory
Bushing defects and their indicators
> Partial breakdowns
> Capacitance
> Partial discharges

> Voids, cracks

> Capacitance
> Partial discharges
center conductor

> Aging by-products, moisture

> Dissipation/power factor

Bushing with capacitive layers and

tap electrode on flange

OMICRON

Page 53

Theory
Capacitance and dissipation/power factor
> Critical indicators of partial breakdowns,
cracks, aging and moisture in bushings
> With online monitoring, these indicators
are measured under real load conditions
> Continuous information about insulation
status allows early corrective action
> Reliable and accurate measurement
results enable comparison with
acknowledged standards

OMICRON

Page 54

Permanent monitoring solution: MONTRANO

MONTRANO monitors critical defect indicators
> Bushing monitoring: Capacitance and dissipation/power factor (DF/PF)
> The classic bushing aging indicators
> Online measurement with lab accuracy

> Partial discharge (PD) in bushings and transformer

> Early detection of defects
> Advanced PD noise suppression and UHFgating

> HV transients monitoring

> Transients are detected directly at the
bushings where disturbances occur

Capacitance
Dissipation/
power factor
OMICRON

Page 55

Partial
discharge
Transient
over-voltages

Permanent monitoring solution: MONTRANO

Benefits
> Continuous assessment of insulation state
> Early detection prevents failures

> Absolute C, DF/PF monitoring ensures

lab accuracy in the field
> Advanced noise suppression for reliable
PD source detection
> Complete HV transient waveform recording
for effective impact assessment
MONTRANO components for continuous
monitoring of transformer dielectric health

OMICRON

Page 56

Permanent monitoring solution: MONTRANO

Benefits
> Web-based data access
>
>
>
>

Fast and easy system status overview

Detailed trend data
Easily adaptable charts and diagrams
Configurable warnings and alerts

> Detailed trend data for modern transformer

health management

OMICRON

Page 57

Permanent monitoring solution: MONTRANO

Customized and extendable system
> One or more transformers can be monitored
> Optional communication with SCADA system
> Fiber-optic communication from acquisition
units to central computer
> Undisturbed communication
> Galvanic isolation
> Up to 4 km (2.48 miles) distance

OMICRON

Page 58

MONTRANO: System components

1

Bushing tap adapters

Reference acquisition unit

UHF sensors

Fibre optic communication

OMS 843 acquisition unit

Central computer and software

1
6

3
4
2

OMICRON

Page 59

MONTRANO: System components

Bushing tap adapters and coupling units
> Synchronously captures signals for
> Capacitance, dissipation/power factor
> Transient over-voltages
> Partial discharge (PD)

> Robust, modular design for different types

of bushings
> Standardized coupling unit
> Customizable tap adapter

> Multiple redundant high-voltage protections

> Built-in temperature sensor
> For temperature compensation of
measured values

> IP65, -40C...+90C (-40F...+194F)

ambient temperature

OMICRON

Page 60

CPL 843 coupling unit

with bushing tap adapter

MONTRANO: System components

UHF sensor
> UHF sensor detects PD signals inside
the transformer tank
> Consists of sensor (antenna) and
signal converter
> Mounting at drain valve or hatch
> Customizable flange adapters optional

UVS 610 drain valve sensor & UHF converter

OMICRON

Page 61

MONTRANO: System components

OMS 843 Transformer acquisition unit
> 4-channel, simultaneous acquisition of data from
the bushing tap adapters and UHF sensor
> Advanced signal processing for capacitance,
DF/PF, transient over-voltages and PD
calculation
> 10 analog inputs for additional measurements
> Humidity
> Ambient temperature
> Optional measurements, e.g. oil pressure

> Fiber-optic communication to central computer

> IP65 enclosure, -30C...+55C (-27.4F...+131F)
ambient temperature
OMS 843 acquisition unit

OMICRON

Page 62

MONTRANO: System components

OMS 843 VT Reference acquisition unit
> Acquisition unit for the reference signal for
capacitance, DF/PF measurement from a set of
voltage transformers (1...3 phases)
> Provides reference signal for absolute
capacitance and DF/PF measurement
> Fiber-optic communication
> Undistorted signal transmission ensures
high-precision measurement
> Galvanic insulation

> System accuracy (at 50/60 Hz, including VT)

> Capacitance: 2pF + dyn. VT error
> DF/PF: 0.05% + dyn. VT error

OMS 843 VT acquisition unit

OMICRON

Page 63

MONTRANO: System components

MCU monitoring control unit
> Fiber optic bus controller
> Transmitter between acquisition units
and central computer
> 3-channel fiber-optic communication
> USB connection to computer
> Integrated in enclosure of central computer or
in separate housing for outdoor mounting
MCU monitoring
control unit

OMICRON

Page 64

MONTRANO: System components

Central computer and monitoring software
> State-of-the art database system for
long-term data storage and retrieval
> Web-based data access &
visualization
> Different user roles / logins:
> Operators:
> Overview of full system setup
> View all data with free configurable
charts and diagrams
> Confirm warnings and alarms

> Administrator: As operator, plus

> Configuration of all monitoring
parameters
> Set and change warning and
alarm rules

OMICRON

Page 65

MONTRANO software
System configuration and control
> Topology views of the monitored assets
> Configuration of all monitoring-related parameters (admin only)
> Configuration of alarm rules (admin only)

OMICRON

Page 66

MONTRANO software
Warnings and alarms
> Defined threshold settings for
alarm notifications
> Real-time display of current
alarm level
> View of events that triggered
an alarm
> Alarm confirmation by operators
> Confirmed alarms change color or
> Are taken out of viewing window
(optional settings)
> All alarm data saved

OMICRON

Page 67

MONTRANO software
Data visualization
> Configurable trend charts show
data for each monitored parameter
> Tooltip pop-up for easy accessable
detailed information of each data
point
> All data records and corresponding
charts can be exported for
reporting

OMICRON

Page 68

Technical specifications
Capacitance, dissipation/power factor
>

Input voltage (at bushing tap) 50 Vrms

Instrument accuracy
> Capacitance
0.8 pF
> Dissipation/power factor 0.01%
System accuracy (at 50/60 Hz including VT)
> Capacitance
2 pF + dyn. VT error
> Dissipation/power factor 0.05% + dyn. VT error

Transient over-voltage measurement

>
>
>
>
>

Input voltage (at bushing tap)

Sampling rate
Rise time
Measurement accuracy
Signal recording depth

Up to 3x Ur
5 MS/s
300 ns (20%...80%)
1.5 %
20 ms...2 s

Partial discharge measurement

>
>
>
>

Frequency range
Sampling rate
Frequency range UHF
System noise

OMICRON

16 kHz...30 MHz
125 MS/s
100 MHz...2000 MHz
< 1 pC

Page 69

References
"Many other systems and techniques had
been tried previously and MONTRANO was
the only system to successfully prove that
the PD was real.
Mr. Colin Feely Plant Maintenance
Engineer, Powercor Australia Ltd.

" With MONTRANO, the results that indicate

possible defects can be correlated quickly.
The ability to see transsient over-voltages at
transformer bushings also offers us new
possibilities to assess the transformer."
Erwin Vgel Voralberg Energienetze
GmbH, Austria

OMICRON

Page 70

Standards
Offline bushing tests
> Capacitance:
> Cigre working group A2.34: Guide for transformer
maintenance presents capacity changes when
one control field layer is short-circuited

> Dissipation/power factor:

1.0

IEC 60137

0.8

max: 0.7%

0.6
0.4
0.2
0.0

typ. new

OIP

typ. new

RIP

Power Factor (%)

Dissipation Factor(%)

> Acceptance level depends on bushing design

1.0

0.4

IEEE C57.19.01
max: 0.85%

0.2

max: 0.5%

typ. new

OIP

typ. new

RIP

0.0
Acceptance level of dielectric losses for bushings of different design

OMICRON

Page 71

No. of
layers

Capacitance
change in %

123

14

7.1

245

30

3.3

420

40

2.5

550

55

1.8

Voltage class and change of capacitance

for condenser type bushings
[Source: Cigre working group A2.34]

0.8
0.6

Voltage
[kV]

Methods
Relative C, DF/PF measurement
> Changes in grid unbalance (phase shifts) can have a misleading impact on
online C, DF/PF measurements when the sum-of-current approach or a
bushing-to-bushing comparison is used to determine the values
> MONTRANO uniquely offers alternative options to reduce the uncertainty
of measurement
Sum of voltages

Relative measurement (bushing-to-bushing comparison)

Example of changes of the sum vector of the 3 phases

of a transformer within 2 days
OMICRON

Page 72

Methods
Benefits of highly accurate C, DF/PF measurement

> Excellent accuracy comparable to offline tests

> Implemented temperature compensation of
measured values
> Long-term stability of measurement

tg in %

> Enables the application of standards for offline

measurement via:

Capacitance (pF)

500
0.8

467 2 pF
measurement

DF (%)

478 pF warning

0.6
Temperature in C

0.4
0.2

400

0.7 % IEC warning

0.27 0.05 %
measurement

RIP
OIP

0.0
Capacitance and DF measurement over 1.75 years, and alarm levels

OMICRON

Page 73

Temperature dependance
of the dissipation factor
[Source: Karl Frey, Micafil]

Methods
Reducing measurement uncertainty
> Grid unbalance impacts are minimized
with the use of a 2nd transformer for
comparison or a VT as reference
> Using a VT as reference, the smallest
uncertainty is achieved

1
Bushing
aging
Reference
aging

Grid
unbalance

Relative
measurement

Highest
uncertainty

OMICRON

Page 74

2
Bushing
aging
Reference
aging
Dual transformer
comparison

3
Bushing
aging

Voltage transformer
reference

Lowest
uncertainty

Methods
Measurement configuration options
3

Relative measurement
(bushing-to-bushing)

Dual transformer
comparison

Lowest
uncertainty

Highest
uncertainty

OMICRON

Absolute measurement
(VT reference)

Page 75

Mathematics behind bushing monitoring

On-line monitoring of dissipation factor and capacitance
1. Reference method
2. Comparison methods
3. Sum-of-current method

Mathematics behind bushing monitoring

Reference method
The reference method can be applied when a group of VTs is available

Mathematics behind bushing monitoring

Reference method dissipation factor
This is the only case when absolute values of the dissipation factor and capacitance can be
obtained. The reference signal taken from the LV side of the VTs, a pure resistive signal (UV),
is compared with the mainly capacitive leakage current measured at the bushing tap (IB) of
the transformer under monitoring. The corresponding vectors of the two signals are rotating in
the same four-quadrant coordinate system. Due to the capacitive nature of it will always lead
in revolution. The angles 1 and 2 between a reference axis and the two vectors are
permanently measured. Computing the difference between 1 and 2 the angle is obtained
and gives the power factor cos. The dissipation angle () and dissipation factor (tg) can be
calculated:

Mathematics behind bushing monitoring

Reference method dissipation factor
&
&
The angle between I B and UVT will decrease with the ageing of the bushings. This fact
involves the increase of the dissipation angle and dissipation factor tg.
The diagnosis of the bushing is done by analyzing the trend of tg and its rate of change.
When tg reaches a predefined level of alarm continuous operation is no longer
recommended. On the other hand, bushings with high values of tg (above a predefined
level) but with stable trend may stay in operation. Another approach derived from practical
experience recommends off-line investigation for bushings whose trend of tg doubles the
value over six months.

Mathematics behind bushing monitoring

Reference method capacitance
The level of the voltage at the LV side of the VTs (ULow) is continuously measured. As the
ratio (rVT) of the VT is known, the values of the voltage on the HV (UHigh) side can be
calculated:

The capacitance CB of a HV bushing is then calculated

IB

2 S f CB U High CB

IB
2 S f U High

The diagnosis of bushings based on their capacitance is done by comparing the values from
on-line monitoring with the values from off-line measurements. The difference between them
gives the capacitance variation C. This quantity (C) has to remain within certain limits. The
example of capacitance change when one control field layer is short-circuited for bushings of
different rated voltage is presented in table 2 [1]. The failure of such a layer does not
necessarily lead the total failure of the bushing but it creates a higher field distribution on the
healthy layers.
Table 2. Voltage class and change of capacitance for condenser type bushings [1]

Voltage in kV
123
245
420
550

No. of layers
14
30
40
55

Change in %
7.1
3.3
2.5
1.8

Mathematics behind bushing monitoring

Reference method capacitance
The calculated capacitance and dissipation factor values obtained based on the reference
signal taken from VTs are stable and not affected by the change in magnitude and phase
angle of the system voltages. Ambient conditions (temperature and hummidity) and operating
temperature (related to the transformer load), bushing surface contamination etc. can affect
the values of dissipation factor and capacitance.
The accuracy of the measurement is given by:
accuracy of the instrument (OMS 843):
< 0.8 pF for capacitance measurement;
0.01 % for dissipation factor measurement.
class of the VT: table below shows the ratio error and the phase displacement for VTs with
different accuracy classes according to [2].

Mathematics behind bushing monitoring

Reference method temperature compensation
The dissipation factor strongly depends on
the temperature of the bushing. Therefore,
temperature correction with respect to the
value of 20C has to be considered

where: is the bushing temperature and
cMat is a correction factor which depends on
the material of the bushings insulation
system (OIP, RIP, RBP-Drysomic and RBPSuperresocel) and has to be calculated for
each bushing separately. The temperature
dependence of the dissipation factor for
different type of bushings is presented below:

Mathematics behind bushing monitoring

Dual Transformer comparison method
The comparison method applies when a group of two sister transformers with the same type
of the bushings are close each other and work in parallel. Here, the HV bushings installed on
the
& same& phase are compared against each other using the capacitive leakage currents
IU and IU measured at their taps
1

Mathematics behind bushing monitoring

Dual Transformer comparison method dissipation factor
The off-line values of the bushings dissipation factors (tgG i ) have to be known (from previous
off-line measurements). These values are used for the calculation of the on-line values of the
dissipation factors ( tgG i' ). A calibration coefficient ( rCAL i ) is introduced in the calculation to
compensate external effects e.g. line delays or asymmetries.

tgG i'

rCAL i tgG i
'

Once the on-line values of tgG U 1 (transformer 1

'
phase U) and tg G U 2(transformer 2 phase U) are
determined, the dissipation angles GU'
and
1
'
G U 2 can be calculated. Computing ' the
difference between them the angle 'G
is
obtained:

'G '

GU'  GU'
2

Mathematics behind bushing monitoring

Dual Transformer comparison method dissipation factor
'

As the bushings are not identical the angle 'G will not be zero. Furthermore, if the two
'
bushings get aged in a similar way, 'G will remain constant. This means, if the rate of ageing
is the same for both bushings, ageing effects cannot be detected.
'

The angle 'G is calculated during an initial

learning phase and is taken as reference value.
This value is compared with the on line
determined values throughout the lifetime of the
''
bushings 'G . Considering that the dissipation
'
'
angles GU1 and G U 2 of the bushings can only
increase (if they decrease in time the bushings
gets better which is not realistic) the diagnosis
can be done as following

'G ' > 'G '' - shows the degradation of the bushing 1
'G ' < 'G '' - shows the degradation of the bushing 2

Mathematics behind bushing monitoring

Dual Transformer comparison method capacitance
The capacitance of HV bushings are calculated as following

IU 1

2 S f CU1off  line U1 U1

IU 1
2 S f CU1off  line

CU1on line

IU1
2 S f U1

where: U1 is the level of the voltage on the HV side and CU 1 off  line is the capacitance of the
bushing from off line measurement (transformer 1 phase U ).
Formula above (left) is applied for each bushing to determine the voltages U1 ... U6. With the
values of the voltage determined using equation above (reight), the on line values of the
capacitance for each bushing are calculated.
A change in the capacitive leakage currents measured at the taps indicates a change in the
capacitance of the bushings. The system voltage variations do not affect the measurement
since both bushings are affected in the same way by these variations. The diagnosis of the
bushings is done based on the difference between the values of the capacitance determined
in off-line and on line mode C. This quantity (C) has to remain within certain limits.

Mathematics behind bushing monitoring

Relative (bushing to bushing) comparison method (one transformer)
When neither VTs nor a second transformer are available, a comparison between the bushing
of the same transformer is performed in order to determine the relative values of the
dissipation factor and capacitance.

Mathematics behind bushing monitoring

Bushing to bushing comparison method
This reference signal capacitive leakage current and
the signals acquired from the other two taps are
digitized and compared against each other considering
the 120 phase shift.
This way, a comparison between the reference vector
(given by the leakage current in the phase U) and the
vectors of the leakage currents in phases V and W is
done. The monitoring system digitizes the three signals
and the angles 1 and 2 are calculated. Based on
these angles, the relative values of the dissipation
factor are obtained.

Mathematics behind bushing monitoring

Sum-of-currents method
The most commonly used method for on line monitoring of capacitive bushings is the socalled sum-of-currents method. It works based on the idea that in a symmetrical three phase
system the sum of the voltage and current vectors is zero. This allows the condition of the
bushings to be determined by vectorially adding the currents measured at the taps of the
bushing under investigation. The resulting vector is called sum current.
If the bushings are
&
identical and the system voltages are balanced, the sum current I 6 will be zero.
Since the bushings& are never identical and the system voltages are never perfectly balanced
the sum current I 6 will not be zero . The resulting vector of the &sum current gives the
reference vector which is compared with the sum currents vectors I 6 calculated along the
life of the transformer. Subtracting the reference sum current vector& from the latest vector of
the sum current a third vector giving the change in the sum current I '6 is obtained. The angle
of this third vector with respect to the reference bushing is used to identify wihich bushing is
causing the change. Once the deteriorated bushing is known the magnitude and the phase of
the change in sum current vector is used to calculate the change in the capacitance and
dissipation factor of this bushing.

Assessment of the 50/60Hz C-Tan Results

Source: Proposal from OMICRON, GBEL and HSP for the Cigre Working Group A2.43 Reliability of HV Bushings

PAGE 90

Commissioning
Standards and recommendations regarding DF
Type
Main insulation

RIP
Resin
impregnated
paper

OIP
Oil
impregnated
paper

RBP
Resin bonded
paper

DF / tan delta (20C,

IEC60137)

< 0.7 %

< 0.7 %

< 1.5 %

PF cos phi (20C, IEEE

C57.19.01)

< 0.85 %

< 0.5 %

<2%

Typical new values

0.3-0.4 %

0.2-0.4 %

0.5-0.6 %

PD (IEC60137) at Um
1.5 Um /3
1.05 Um /3

< 10 pC
< 5 pC
< 5 pC

< 10 pC
< 5 pC
< 5 pC

< 300 pC

PAGE 91

Assessment and Interpretation

Indication

RBP

OIP

RIP

increase of capacitance

oil in cracks
or partial breakdowns

partial breakdowns

partial breakdowns

high dissipation factor

partial breakdowns;
insulator surface wet or
dirty (clean the insulator);
ageing of the inner
insulation;
water in the inner
insulation;

partial breakdowns;
insulator surface wet or
dirty (clean the insulator);
ageing of the inner
insulation;
water in the inner
insulation;

partial breakdowns;
insulator surface wet or dirty
(clean the insulator);
ageing of the inner
insulation;
water in the inner insulation;

dissipation factor is
decreasing with
increasing voltage

bad potential connections;

partial breakdowns

bad potential connections;

partial breakdowns

bad potential connections;

partial breakdowns

dissipation factor is
strongly increasing with
increasing temperature

high moisture in the

insulation;
high degree of ageing

high moisture in the

insulation;
high degree of ageing

high moisture in the

insulation;
high degree of ageing

normal, if constant

Discharges produce
gasses;
Errosion of the cellulose;
production of x-wax

partial breakdowns;
cracks or voids after
electrical or mechanical
stress;

partial discharges

Methods
Partial discharge (PD) detection
> PD detection is the early indicator of defects in dielectric insulation
> In typical field environments of transformers, electrical PD detection
suffers from a high noise level
> Originated from outside the transformer, like corona

> MONTRANO monitors PD at the bushing taps and inside the

transformer tank, using a UHF sensor
> The sensor inside the tank is not affected by outside noise

> Synchronous multi-channel recording enables UHF gating

> Using the UHF signals to distinguish between PD events inside and
outside the transformer at the bushing tap sensors

> Further advanced noise suppression technologies are available:

> 3-Phase Amplitude Relation Diagram (3PARD)
> 3-Center Frequency Relation Diagram (3CFRD)

OMICRON

Page 93

Methods
PD denoising and source separation
Phase C

Phase C

Phase B
Phase B
Channel 2
3-band filter

Phase A

Phase A

Channel 1
3-band filter

PD diagrams of 3 phases
without separation

OMICRON

Channel 3
3-band filter

Measurements related to each

other in a 3PARD diagram

Page 94

Separated PD activity:
located in phase C

Methods
Recording of transient over-voltages
> Detection directly at bushings
> Where the dielectric stress occurs

> Recording of full signal shape at all 3 phases

> Supports IEEE compliant export format (COMTRADE C37.111-1999)
2.0e+005

1.5e+005
1.5e+005

1.0e+005
1.0e+005

5.0e+005

Voltage in kV

Voltage in kV

5.0e+005

0.0e+005

-5.0e+005

-1.0e+005

0.0e+005

-5.0e+005

-1.0e+005

-1.5e+005

-1.5e+005

0.095

0.100

0.105

t in ms

0.110

0.115

0.085

Different examples of transients recorded at bushings of a power transformer

OMICRON

Page 95

0.090

0.095

0.100

t in ms

0.105

0.110

0.115

0.120

0.125

Topics
> Diagnostic Monitoring Systems
> Permanent and Temporary PD Monitoring System for Rotating
Machines
> PD, C, TanDelta, transient Monitoring System for Transformers
> Permanent and Temporary PD Monitoring System for HV Cables

Page 96

Monitoring of short- and long-link HV cables

Steffen Moeck
Monitoring Project Sales
OMICRON Energy Solutions GmbH, Berlin

OMICRON

March 5, 2015

On-line monitoring on HV cable systems

WHERE?
> Terminations
> Joints
WHAT?
Partial Discharge

OMICRON

Seite 98

Insulation status assessment in HV cable systems

Why on-line PD monitoring ?
> Provides on-line condition assessment
of insulation in HV cables and cable
accessories (terminations and joints)
during standard operating conditions
> Detects insulation defects early to
prevent failure
> Supports cost effective, condition-based
maintenance to extend service life
> Powerful analysis software and server
architecture enables long-term data storage,
visualization & evaluation

OMICRON

Page 99

Temporary and permanent monitoring solutions

Temporary monitoring
system

Permanent monitoring system

OMICRON

Page 100

Permanent PD monitoring system

for HV cable terminations and joints

OMICRON

10 March 2015

Permanent monitoring solution

Condition monitoring to extend
HV cable system life
> Continuous assessment of insulation state
in HV cables, joints and terminations
> Early detection prevents failures

> Advanced PD source separation by

synchronous multi-channel (3PARD) and
multi-spectral (3CFRD) evaluation
> Diverse HFCT sensors for installation at
groundings or cross-bonding boxes
> Inductive power supply (IPS) for offside
supply of units
> Optional oil pressure monitoring and
distributed temperature sensing (DTS)
can be added
> Web-based data access, management
and visualization
OMICRON

Page 102

Continuous monitoring system

for high-voltage cable systems

System configuration
Extendable system for long and short link
HV cable systems

Acquisition unit
Fiber optic cable

Inductive
power supply

HF current trasformer

OMICRON

Page 103

Central computer &

monitoring software

System components
1

Current transformer

Fiber optic communication

Acquisition unit

Fibre optic bus

Inductive power supply

Central computer and software

6
2

1
2

4
1

OMICRON

Page 104

System components
MCT 120 high-frequency current transformer (HFCT)

> Designed to pick up PD signals

from readily accessible positions
and at a safe distance
> Easy installation on ground connections
> Ample clearance for large cables
> Sealed IP67 casing protects sensor
from adverse conditions
> Bandwidth
35kHz-40MHz,

0 mm gap

180kHz-40MHz,

0.22 mm gap

290kHz-40MHz,

0.44 mm gap

380kHz-40MHz,

0.66 mm gap
MCT 120 HFCT

OMICRON

Seite 105

System components
MCT 120 high-frequency current transformer (HFCT)
> Fixation to wall or frame

> Fixation at the cable

Cable Clambing

Universal
Page 106

System components
UCS1 high-frequency UHF sensor (option)
> Measures PD in high frequency
UHF range
> Designed specifically for PD detection
at HV cable terminations
> Ensures sensitive measurements
in noisy environments
> Galvanically isolated signal output
prevents induced disturbance currents

UCS1 UHF sensor

OMICRON

Seite 107

System components
Data acquisition unit
> Synchronous multi-channel PD data acquisition
> Advanced technology like 3PARD and 3CFRD for PD source separation
> Optical isolation between partial discharge acquisition units and server
using fiber optic cable

OMICRON

Page 108

System components
IPS 820 inductive power supply
> Allows operating the acquisition units offside
> Ideal for cable tunnel applications where copper
wiring has to be avoided and voltage sources are
not available
> Electronic elements are filtered and optimized to
prevent measurement disturbance

IPS 820 inductive power

supply and cable

OMICRON

Page 109

Software
Central computer and monitoring software
> State-of-the art database system for longterm data storage and retrieval
> Web-based data access & visualization
> Different user roles/logins:
> Operators:
> Overview of full system setup
> View all data with free configurable charts
and diagrams
> Confirm warnings and alarms

> Administrator: As operator, plus

> Configuration of all monitoring parameters
> Set and change warning and alarm rules

OMICRON

Page 110

OMS 605
Temporary PD monitoring for HV cable terminations and joints

OMICRON

10 March 2015

Temporary monitoring solution: OMS 605

Why periodic PD monitoring with OMS 605?
> Partial discharges (PD) indicate damaged or aging insulation
> Detects insulation defects early to prevent asset failure
> Indicates which HV assets need to be monitored on a continuous basis
> Supports cost effective, condition-based maintenance to extend asset life

OMICRON

Page 112

Temporary monitoring solution: OMS 605

Benefits
> 3 channels for synchronized
PD data acquisition
> Plus one channel for additional
acquisition unit
> Wheeled case enables easy transport,
to different monitoring locations
> Rugged measurement system
for industrial environments (IP65)

OMS 605 portable PD monitoring system

for high-voltage assets

> Simple to use plug-and-play modules

for fast system set-up
> Compatible to a variety of sensors
> Powerful software for PD analysis,
trending and data visualization
OMICRON

Seite 113

Temporary monitoring solution: OMS 605

Portable system for one group of cable terminations...

OMS 605
Notebook
with software

HFCT
sensors

OMICRON

Seite 114

Temporary monitoring solution: OMS 605

... or portable system for multiple groups of cable accessories

Notebook
with software

OMS 605

HFCT
sensors

OMICRON

Seite 115

OMS 605: System components

MCT 120 high-frequency current transformer (HFCT)

> Designed to pick up PD signals

from readily accessible positions
and at a safe distance
> Easy installation on ground connections
> Ample clearance for large cables
> Sealed IP67 casing protects sensor
from adverse conditions
> Bandwidth
35kHz-40MHz,

0 mm gap

180kHz-40MHz,

0.22 mm gap

290kHz-40MHz,

0.44 mm gap

380kHz-40MHz,

0.66 mm gap
MCT 120 HFCT

OMICRON

Seite 116

OMS 605: System components

MCT 120 high-frequency current transformer (HFCT)
> Fixation to wall or frame

> Fixation at the cable

Cable Clambing

Universal
Page 117

OMS 605: System components

UHF extension kit (optional)
> Protected box contains:
MPD 600 + MPDD 600 + UHF 620
> Used for connecting the OMS 605 to
high-frequency UHF sensor(s) at
cable terminations
> Can be used as single UHF-channel
or as a 4th UHF channel together
with the 3-channel OMS 605

UHF extension kit

OMICRON

Seite 118

OMS 605: System components

UCS1 high-frequency UHF sensor (option)
> Measures PD in high frequency
UHF range
> Designed specifically for PD detection
at HV cable terminations
> Ensures sensitive measurements
in noisy environments
> Galvanically isolated signal output
prevents induced disturbance currents

UCS1 UHF sensor

OMICRON

Seite 119

OMS 605: System components

Mobile monitoring software
1

Set up a monitoring session

> Select an existing asset template

or create a new one
> Define monitoring duration and
frequency
> Set warning and alarm thresholds
> Import settings from previous
monitoring sessions or diagnostic
tests

OMICRON

Seite 120

OMS 605: System components

Mobile monitoring software
1

Start a monitoring session

> View running monitoring session

> Capture pre-analyzed PD trend
data
> Record 30-second data streams
to be analyzed using the expert
PD software

OMICRON

Seite 121

OMS 605: System components

Mobile monitoring software
2

View PD trend data from

completed sessions

> Warning / alarm indication notifies

you when user-defined thresholds
have been exceeded
> Scroll over data points to view
time and value of PD activity
> Click on data points to display
PRPD and 3PARD images
> Trend data can be easily exported
in a variety of formats for reporting

OMICRON

Seite 122

OMS 605: System components

Mobile monitoring software
2

View PD trend data from

completed sessions

> PRPD (Phase-resolved PD)

and 3PARD (3-Phase
Ampliture Relation Diagram)
images show overlapping
and separated PD activity

PRPD

> Trend data can be easily

exported in a variety of
formats for reporting

3PARD
OMICRON

Seite 123

OMS 605: System components

Expert PD analysis software
3

Real-time diagnosis of PD data

> Various filters can be applied to

see more detail of PD activity
> Resulting information helps you to
reliably assess insulation
condition status
> Data can be exported for reporting
in a variety of file formats (i.e.
Excel, MATLAB)

OMICRON

Seite 124

Presentation 25.1

Monitoring of Rotating Electrical Machines

Maik Koch, FKH, Switzerland
Steffen Kurz, Omicron Energy Solutions, Germany

Abstract
This
paper
describes
general
design
considerations, the system layout and practical
experiences with on-line monitoring of electrical
apparatus in large thermal power plants. The
availability of electrical equipment in power plants
is of utmost importance, since outages can easily
lead to financial loss of millions of Euros, even per
day. Therefore operators of power plants pay large
attention to the system health of their apparatus,
particularly these in the functional chain of power
generation, as e.g. generators and generator stepup transformers. On-line monitoring appears to be
an important tool to provide reliable status
information, detect failures early and utilize the full
life time.
For identifying possibly unreliable components,
general failure statistics provide first input. For
example in the work of Cigr WG A2.37
"Transformer Reliability Survey", most failures
have been attributed to the dielectric of power
transformers. The same failure mode applies for
generators. Experiences with specific grid
conditions and particular apparatus designs are
another input for selecting appropriate monitoring
systems. Finally, economic calculations prescribe
the frame of the extent of the applied technology.
For large turbo generators, it was decided to
monitor the temperature in the slots of the stator
and the partial discharges of the stator. The
measurement of partial discharges is carried out in
a frequency range of several 100 kHz to a few
MHz for being more sensitive towards signals
coming from the inside of the stator windings.
Rotating electrical machines generally have
several sources of PD, therefore source separation
is essential to discriminate between noise, PD with
low failure probability and dangerous PD. In the
context of this work, amplitude-based impulse
correlations were most effective (3 phase
amplitude relationship diagram - 3PARD). Alarms
are based on levels as well as on rate of change.
PD pattern classification is applied for a basic risk
assessment.
For the large generator step-up transformers it was
decided to monitor temperatures, dissolved gases
in oil (DGA), bushing capacitance and dissipation
factor, partial discharges and transient overvoltages. Transient over-voltages where suspect to
be the reason for a number of breakdowns in the
last decades, so their monitoring with true
waveform recording was of high importance. Actual
measurements show the interaction between the
transformer and the grid. For monitoring partial

discharges, the traditional measurement of

electrical PD pulses at the bushing taps is
combined with UHF measurements using drain
valve sensors. This is necessary for discriminating
between internal, potentially dangerous PD and
external corona. For determining the state and
ageing condition of the bushings, different methods
are used in comparison: sum of current vectors
from three phases of one transformer, bushing-tobushing comparison and the voltage transformer
(VT) reference method. Highest overall system
accuracy was achieved with the VT reference
method, where the secondary voltages of VT's are
used as reference for calculating capacitance and
dissipation factor. The typical over-all system
accuracy is 2 pF for capacitance and 0.5 % for
dissipation factor. All measurement functions are
illustrated with practical experiences and case
studies which help to prove the benefit of
monitoring of electrical apparatus in large power
plants.

Condition Monitoring and

Diagnostics in the Context of
Energy Production
The availability of electrical equipment in power
plants is of utmost importance, where outages can
easily lead to financial loss of millions of Euros.
Therefore operators of power plants pay large
attention to the system health of their apparatus,
particularly these in the functional chain of power
generation, as e.g. generators and generator stepup transformers. In this application scenario,
condition monitoring and diagnostics promises to
provide reliable status information, detect failures
early and utilize the full life time of equipment.
For many decades, power equipment users have
sought ways to assess the general condition and
identify specific problems of their assets.
Conclusively, diagnostic tests have been
developed, which are applied in the de-energized
status. In the last years, a sophisticated means has
evolved for collecting a great deal of diagnostic
information while the equipment is in service. While
periodic off-line diagnostic tests still play the
dominating role in condition assessment,
continuous or on- line monitoring promise to
have the potential to overcome some of the
fundamental limitations of off-line tests:
- Continuous measurement for having
reliable measurement data, reducing the
effect of "outliers" and for continuous
observation of the equipment's condition;

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

Presentation 25.2

Early diagnosis of initiating failures for

scheduling maintenance actions, therewith
supporting condition based maintenance
schemes;
- Knowledge of the equipment's historical
use for fully utilizing the life span in the
context of asset life management.
- On-line monitoring is apparently becoming
an essential feature of the "smart" electric
utility systems of the future.
The context of large power plants requires specific
features of instruments used for condition
monitoring and diagnostics:
- For their costly and reliable assets, power
plant operators require extreme reliability of
the monitoring instruments.
- The question "How to get real value from
continuous on-line monitoring systems?" [1]
indeed needs an answer. The integration of
monitoring results into existing SCADA
systems requires a high abstraction level
and sophisticated analysis algorithms.
- Outages need to be avoided. Therefore not
only a warning system, but also diagnostic
functions are required.
A general first step into condition monitoring is a
strategic look into failure statistics for identifying
possibly unreliable components. For example in
the work of Cigr WG A2.37 "Transformer
Reliability Survey", most failures are of dielectric
nature [2]. The same failure mode applies for
generators [3]. Based on this, monitoring of the
dielectric integrity seems to be of highest
importance. Experiences with specific grid
conditions and particular apparatus designs are
another input for selecting appropriate monitoring
systems. Finally, economic calculations prescribe
the frame of the extent of the applied technology.
-

Thermal;
17%

Mechanical
; 27%

Unknown;
14%
Chemical; 3%
Thermal;
11%

Electrical;
15%

Dielectric;
56%

Dielectric;
37%

Mechanical
; 20%

Condition Monitoring of Large

Turbo Generators
Generator On-line Monitoring
Maintenance, inspection and the majority of tests
of large hydrogen cooled turbo-generators is only
possible within planned outages of the power plant
and is subject to significant cost pressure that have
resulted in decisions to reduce maintenance and
inspection frequency and accept greater operating
risk. This trend can be observed world-wide [4]. In
contrast to off-line tests, in-service diagnostic tests
can access the generator integrity under real
service conditions and allow for planning outages
on a long term, if necessary. Table 1 lists inservice or on-line applied tests in the context of this
work.
Table 1: Diagnostic measurements for large hydrogencooled turbo generators during operation (on-line
monitoring)

Component
Stator
winding
Rotor
bearings
Cooling
system

Monitored Parameter
Partial discharge monitoring
Slot temperature monitoring
Vibration monitoring
Hydrogen dew-point and
hydrogen purity monitoring

Experiences with Partial Discharge

Monitoring
Based on a number of failure statistics [3], partial
discharges are a dominant root cause for insulation
breakdowns;
therefore
their
continuous
observation received high priority. Partial
discharges result from stress of the insulation as
caused by temperature cycles, the applied
electrical stress, aggressive chemical substances,
moisture, contamination and mechanical forces.
The stator insulation of rotating machines generally
exhibits a number of sources of partial discharges;
there is no PD-free stator insulation. The
fundamental challenge of on-line monitoring is to
discriminate between discharges with high,
medium and low risk and external disturbances.
During in-service monitoring of partial discharges,
a number of these PD sources will be active at the
same time. In Figure 2 these are exemplarily
disturbances without any relevance for the
insulation
condition,
end-winding
surface
discharges with medium risk and delamination of
tape layers with high risk. As these single sources
cannot be measured separately, their specific PD
patterns are shown superimposed in a phase
resolved PD event representation (PRPD
representation). The apparent charge is then a
result of the superposition of all PD sources. An
alarm based on such superposition of different PD

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

Presentation 25.3

sources will not be able to separate the critical

source, but may lead to false positives.
1.38 nC

Disturbances

Transformer Tests during Outages

200 pC

Today a large variety of off-line applied diagnostic

tests for transformers is available. Table 2 lists the
tests applied in the context of this work.

Background Noise and Disturbances asynchronous with AC

medium

E1
900 pC

Table 2 Diagnostic tests for large generator step-up

transformers during outages

Inner PD + End winding surface

discharge - contamination

Alarm

Q
high

Condition Monitoring of Large

Power Transformers

S2
500 pC

Component
Insulation
system

Delamination of tape layers - bigger

symmetrical inner micro void

time

Figure 1: Superposition of various PD and noise sources

during on-line monitoring of partial discharges

To separate between various PD sources,

synchronous multi-channel measurement is
applied [5]. In the context of this work, three
coupling capacitors of 1.1 nF each are mounted
into the generator busbar. The 3 phase amplitude
relationship diagram (3PARD or star diagram)
visualizes the relation among amplitudes of a
single PD pulse in one phase and its crosstalk
generated signals in the other two phases. By
repetition of this procedure for a large number of
PD pulses, PD sources within the test object as
well as outer noise appear as clearly
distinguishable clusters of dots in a 3PARD
diagram. Therefore, the 3PARD diagram shows
different pulse-type sources in separable clusters.
Each cluster can be selected individually and the
pattern displayed in a PRPD. No other source is
included in the pattern. Figure 3 illustrates the PD
source separation by the 3PARD principle as well
as the back-transformation of individual clusters
into the PRPD representation. In the context of this
work, PD expert analysis is performed in the
frequency of 1 month. New clusters are separated
and individually evaluated via remote access to the
monitoring system. Automatic PD source
separation and pattern classification is the next
step of on-line PD monitoring, [6].

Windings

Bushings
OLTC

Frequency
6 or 12 months

Dissolved gas
analysis
Furanes

6 or 12 months

Dielectric response
analysis
Winding resistance
Transmission ratio
Short circuit
impedance
No load current
Frequency
Response Analysis
FRA
Capacitance and
dissipation factor
Dynamic resistance

2 years / After
indication
After indication
8 years
8 years
8 years
8 years
8 years / after
indication
4 years
4 years

Transformer On-line Monitoring

Based on international and local failure statistics,
the on-line monitoring techniques of Table 3
received the highest priority from the authors of
this work. The selection is based on discussions
with other power plant operators, own experiences
and available technology.
Table 3: Diagnostic measurements for generator step-up
transformers during operation (on-line monitoring)

Component
Insulation
system

Bushings
OLTC
Figure 2: An example of separation of PD sources by
3PARD

Diagnostic Test
Oil analysis

Monitored Parameter
Oil temperature
Dissolved gas analysis (gas
selective multi-gas systems)
Moisture in oil
Partial discharges
Transient over-voltages
Capacitance and dissipation factor
Motor current analysis

Experiences with DGA Monitoring

The analysis of dissolved gases in oil (DGA) is one
of the most powerful tools for diagnostics of oil-

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

Presentation 25.4

C2H2
CH4
H2

C2H4
CO
AmbientTemp

120

C2H6
CO2
OilTemp

100

Gas concentration / ppm

1000

80
100
60
10
40
1

Temperature / C

10000

20

0,1

0
25-6

30-6

5-7

10-7

15-7

20-7

Date

25-7

Figure 3: On-line monitoring data of dissolved gases and

temperature for a period of 1 month, showing strong
increase of most hydrocarbons after 03. July

Figure 4 shows the individual gases measured online over a time period of one month. At [Link],
the hydrocarbons strongly increased. DGA
laboratory measurements from oil samples
confirmed the on-line readings. Various analysis
schemes were applied, pointing towards a local hot
spot, although not congruently. The very specific
conclusion of the Rogers assessment rule finally
agreed with the actual failure. In discussion with
the transformer manufacturer the possible location
of the hot spot was identified and later, after
opening the transformer, confirmed. A short circuit
between the tank and one field grading electrode
of the winding caused circular currents which lead
to over-heating and burned the oil-paper insulation.
For operation of a power plant, such early
detection of an ongoing failure is of outmost
importance. This allowed early planning of the
outage, searching for a replacement of the
transformer and performing the exchange in less
than 14 days. An unplanned outage e.g. based on
Buchholz tripping might easily result in an outage
of months duration with large consequent financial
losses.

Experiences with Bushing Monitoring

Since bushings are subjected to high dielectric and
thermal stresses; bushing failures are one of the
root causes of forced outages and transformer
failures. These are even connected with bushing
explosions. For monitoring the dielectric integrity of
bushings, three measurement parameters cover all
important bushing failure modes: bushing
capacitance, dissipation factor (tangent delta) and
partial discharges. In the context of this work a
combined measurement of these parameters is

1
40
0.9
35
0.8 IEC60137 max: 0,7 %
30
0.7
25
0.6
0.5
20
0.4
15
U-phase W-phase
0.3
10
0.2
V-phase
V
h
5
0.1
0
00
5
10
time / days 15

Figure 4: On-line measurement of bushing dissipation

factor

The needed accuracy for reliable

detection of failures is in the range of 5 pF
for capacitance and 0.1 % for dissipation
factor. For this measurement, generally a
reference signal is necessary but in the on-line
application, the phase voltage is not directly
available. Therefore various approaches for
obtaining such reference signal were investigated
with the result, that the secondary voltage of
voltage transformers provides the highest accuracy
[6]. In various field installations a typical overall
system accuracy (including voltage transformers)
of 2 pF for capacitance and 0.05 % for
dissipation factor was reached, even without
further data processing like averaging. In case
voltage transformers are not available, bushings of
a second transformer provide the reference signal.
This principle was used for the results shown in
Figure 5. The dissipation factor of three bushings
vs. time shows a week dependence on
temperature, but even without automatic
temperature compensation, the measurement is
stable and accurate. Temperature was measured
in the bushing flange. Between days 8 and 10,
heavy rain lead to a drop in temperature and made
the bushing surfaces wet, resulting in the visible
slight increase of dissipation factor.
The accuracy of this system can be compared to
accurate off-line measurements, consequently
standards from off-line tests like IEC 60137 are
implemented as warning and alarm levels. With
this solution, on-line monitoring is not only
trending, but provides the accuracy for clear good /
bad evaluation and maintenance decisions.

Experiences with Transient OverVoltage Recording

Transient over-voltages originate from events in
the HV grid and may lead to pre-mature insulation
ageing. After the experiences of the power plant
operators, over-voltages are likely the reason for a
historical
catastrophic
transformer
failure.
However, there are extremely limited possibilities
to regularly monitor transient over-voltages and to

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

temperature in C

performed by one instrument connected to the

bushing taps.
dissipation factor in %

filled equipment because of the rich operation

experience
and
the
mature
technical
implementations. Generator step-up transformers
belong to the most expensive equipment in power
plants; therefore gas selective multi-gas sensors
are generally preferred to sum-gas sensors. Using
the single gases, advanced interpretation schemes
like the Doernenburg ratios and the Duval triangles
are applied.

Presentation 25.5

make a full recording of the time signal up so

several MHz. In the context of this paper, transient
over-voltages are recorded via the bushing tap.
Measurements of the transmission behaviour of
bushings proved their linearly capacitive behaviour
until at least 16 MHz. Figure 6 shows transient
over-voltages measured at two parallel power
transformers at the 400 kV side, where oscillations
of 10 kHz and superimposed oscillations of 600 Hz
frequency are visible. The reason for these
oscillations is under investigation.

acceptance criteria, this transformer seems to

develop a failure. Actually, this high PD activity is
originated by corona within the substation. After
our
experiences,
solely
electrical
PD
measurements at the bushing taps are useless for
observing the transformers insulation health.
QIEC in nC

10

Ph. V

1
Ph. U

0.1

400

W Phase
0.01
0

10

t in ms

20

Figure 6: PRDPD pattern of electrical on-line

measurement at bushing taps (left) and corresponding
3PARD (right).

0
V Phase
-200

100
-60

12

t in ms

18

U in V

U Phase

P in dBm

Voltage in kV

200

-400

Ph. W

-100

Figure 5: Transient over-voltages measured at 400 kV

bushings

-120

0.001

Experiences with Partial Discharge

Monitoring
Partial discharge (PD) is a localized dielectric
breakdown of a small portion of a solid electrical
insulation system. Since partial discharges are
early indicators of incipient faults, their on-line
observation is of prominent interest. Acceptance
criteria in the factory test of power transformers are
an apparent charge of less than 5 pC for bushings
(IEC 60137) and less than 300 pC for power
transformers (IEC 60076-3).
For on-line measurement of partial discharges, two
measurement principles are popular: (1) detection
of electrical signals at the bushing taps and (2)
detection of electromagnetic waves with UHF
sensors. On-line partial discharge measurements
are subject to an immanent thread: the
discrimination between external corona and
internal,
"true"
partial
discharges.
Solely
measurements at the bushing taps struggle with
this discrimination.
QIEC in nC

10

Ph. V

1
Ph. U

0.1
0.01
0

10

t in ms

Ph. W

20

Figure 7 (left) depicts phase resolved partial

discharge patterns (PRPD) for three phases,
measured at the bushing taps. The apparent
charge after IEC ranges from 8 to 20 nC,
depending on weather conditions. Recalling the

10

t in ms

20

500
1000
f in MHz

Figure 7. UHF PD measured using a drain valve sensor:

Phase resolved pattern showing only instrument noise
(left) and frequency sweep with CW broadcast stations
(right).

While electro-magnetic disturbances pollute the

environment outside the transformer tank, the
transformer tank is free of discharges. Drain valve
or hatch type UHF sensors inside of the
transformer tank pick up internal electromagnetic
impulses from partial discharges and are at the
same time immune against external corona. The
combination of signals in the UHF range with
electrical signals from the bushing tap promises to
provide a high sensitivity together with suppression
of external noise like corona. Figure 8 proves this
with the UHF PRPD (left) and the UHF frequency
sweep (right). The UHF PRPD shows only
instrument noise of very low amplitude, i.e. 3 V. In
the frequency sweep, measured while tuning the
UHF receiver from 180 to 2000 MHz, only very
week signals from mobile radio transmitters are
visible e.g. at 890 MHz. These conditions allow for
a sensitive and noise-robust detection of inner
partial discharges. In this context, the combination
with electrical measurements at the bushing taps
provides additionally phase information and often a
higher sensitivity to internal defects than sole UHF
measurements.
Additionally,
software-based
correlation algorithms like the 3PARD in Figure 3
can help to separate various PD or corona
sources.

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

Presentation 25.6

System Design for Combined

Monitoring of Generator and
Transformer
This section describes one example of a real field
installation. At a large brown coal power plant, one
900 MW block was equipped with a monitoring
system for the parameters:
- Partial discharges at the 900 MW generator
measured with capacitive couplers
- Partial discharges at two 1.1 GVA
transformers, including electrical and UHF
PD detection
- Transient
over-voltages
at
both
transformers
- Capacitance and dissipation factor of both
transformer bushings
- For this installation, DGA was monitored
using an independent instrument and not
integrated in the structure of Figure 9.
UMTS

Voltage Transformers

OMS843
- C/DF
(Reference)

Generator

PDM600

ca. 10 m

OMS843
- PD
- Transients
- C/DF

[1] P. Jarman et al.: "Recommendations for condition

monitoring and condition assessment facilities for
transformers", Cigr WG A2.27, Technical Brochure
343, on-line available at [Link]
[2] S. Tenbohlen et al.: "Transformer Reliability
Survey", Cigr WG A2.37, preliminary tutorial
presented at the ETG Diagnostik elektrischer
Betriebsmittel, Fulda 2012
[3] Brtsch et al. "Insulation Failure Mechanisms of
Power Generators", DEIS July/August 2008
[4] J. Linton et al.: "Generator Maintenance, Inspection
and Test Programmes", Cigr WG A1.07, Technical
Brochure 386, on-line available at [Link]
[5] W. Koltunowicz and R. Plath, "Synchronous multichannel PD measurements", IEEE Transactions on
Dielectrics and Electrical Insulation, vol.15, no.6,
December 2008, pp. 1715-1723.
[6] A. Obralic, W. Koltunowicz and A. Belkov:
"Automatisches Mustererkennungsverfahren bei TEMonitoring an rotierenden elektrischen Maschinen",
ETG Diagnostik elektrischer Betriebsmittel, Fulda
2012
[7] M. Koch, M. Krger, M. Lukas " Ein neues
Verfahren
zum
Online-Monitoring
von
Durchfhrungen, Transienten und Teilentladungen
an Transformatoren", ETG Diagnostik elektrischer
Betriebsmittel, Fulda 2012

About the Author

OMS843
- PD
- Transients
- C/DF

Transformer 1

Transformer 2
UHF620 + UVS

Bibliography

UHF620 + UVS

Figure 8: System design for combined monitoring of

generator and two transformers at a 900 MW power unit
of a lignite fired power plant

Figure 9 illustrates the design of the monitoring

system. In this case, three reference principles for
calculating capacitance and dissipation factor were
compared; the sum of currents, phase-to-phase
comparison and VT reference. The power plant
operator had particular interest to monitor transient
over-voltages in the grid, as historically transformer
breakdowns were attributed to lightning strokes.
For
PD monitoring,
both the electrical
measurement at the bushing taps and the UHF
method were combined. All monitoring acquisition
units are directly mounted at the HV equipment
and are synchronized via fibre-optic cables. A
central computer stores and evaluates the data,
generates warnings and alarms and provides a
web interface. Via common web browsers, users
can log into the system, access the data, change
alarms etc.. Measurement results of this system
are demonstrated in Figures 5-8.

Maik Koch leads the Insulation

Material Laboratory at the FKH
(Expert Commission for HV
engineering and testing) in
Switzerland. He studied electrical
power engineering at various
German
universities
and
graduated as a Ph.D. at the
University of Stuttgart in Germany in 2008. In
2007, he joined Omicron electronics, Austria,
where he lead the Product Management. He joined
FKH in 2013. His field of expertise is condition
assessment of HV assets by electrical, chemical
and dielectric analysis methods. He wrote more
than 70 scientific papers and contributes to
working groups of VDE, CIGRE, IEC and IEEE
dealing with subjects such as HV testing and
diagnostics, insulation ageing and on-line
monitoring.

OMICRON electronics Asia Limited 2013 Conference on Electrical Power Equipment Diagnostics

RETURN OF EXPERIENCE FROM CONTINUOUS

PD MONITORING OF ROTATING MACHINES
*

L.V. Badicu , W. Koltunowicz, M. Koch and A. Piccolo

OMICRON Energy Solutions GmbH, Lorenzweg 5, Berlin 12099, Germany
*Email: [Link]@[Link]
Abstract: The experience from continuous partial discharge monitoring to assess the
condition of generator stator winding insulation is presented. The case studies are
related to monitoring of hydro and turbo generators of different rated power from 140
MVA to 1160 MVA. The advanced features for elimination of disturbances and for
separation of different types of insulation defects based on synchronous, multichannel and multi-frequency techniques are shown. The examples of data evaluation
are described in detail and the use of the automated PD pattern recognition system is
also discussed.
1

INTRODUCTION

On-line, continuous partial discharge (PD)

monitoring has been used for decades to assess
the condition of a generator stator winding
insulation system. Modern digital monitoring
equipment, with advanced hardware and software
capabilities, has allowed improving insulation
diagnostics by means of PD analysis. The
determination of type of the defect and its severity
is easier at machine normal operating conditions
and the results of the data evaluation are more
accurate. This helps maintenance engineers set up
an effective condition based maintenance program
[1].
A previous paper by the authors described in
details the elements of a modern PD diagnostic
system [2]. This paper deals with the aspect of
interpretation of PD monitoring data, coming from
practical cases studies. Four cases related to
machines with different insulation technology as
well as different rated voltages and powers are
considered:
turbo generator of 1160 MVA and 27 kV
turbo generator of 436 MVA and 20 kV
turbo generator of 180 MVA and 16.5 kV
hydro generators of 140 MVA and 16.5 kV
The PD data are synchronously acquired from all
three phases and multi-channel techniques for
noise elimination and PD source separation are
applied to make accurate insulation conditions
assessment. The automated pattern recognition
system is also used to make the analysis faster
and with limited need of human expert
involvement.
2

SYSTEM CONCEPT

The concept of a modern on-line continuous

monitoring system is presented in Figure 1 [2]. The
signals from PD sensors are synchronously
acquired in a three-channel data acquisition unit,

which performs pre-processing of the raw data to

remove disturbances and to determine PD
characteristics. The monitoring data can be
acquired in two modes, permanent and periodic,
which are flexible and can be defined by the user.
During normal operation only data from periodic
mode are stored to avoid overloading of the data
base. Example of data from periodic mode is
shown in the PD trend diagram in Figure 2. For
each measurement, scalar values (apparent
charge, repetition rate of the pulses, AC voltage
phase and the absolute time) and phase resolved
partial discharge (PRPD) patterns are saved.
In case of an alarm (threshold violation), data
acquired in permanent mode will be automatically
stored and displayed in the trend diagram. Thus,
only the most relevant monitoring data is kept for
analysis while redundant information is discarded.

Figure 1: Concept of continuous PD monitoring [1]

3

CASE STUDY:
1160MVA, 27KV

TURBO

GENERATOR

The 1160 MVA, 27 kV turbo generator was put in

operation in 1998 and in 2012 was taken out of
service for regular maintenance.

Figure 2: Three phase PD trend diagram

The on-line PD monitoring system was installed
during the outage and it started immediately
collecting data when the generator was put back
into operation. It can be seen in the three phase
PD trend diagram (Figure 2) that the PD level
decreases after start up. After the conditioning time
of two months a data processing was performed
with the support of an automated pattern
classification system. The accuracy of automatic
data evaluation depends on the efficiency of
preliminary separation of different PD sources and
on effective suppression of disturbances and
background noise. To achieve this, synchronous
multi-channel evaluation techniques were applied
[3].
The synchronous multi-channel PD evaluation
technique (3PARD diagram) visualizes the
relationship between the amplitudes of a single PD
pulse in one phase and its crosstalk generated
signals in the other two phases. By repetition of
this procedure for a large number of PD pulses, PD
sources within the test object, as well as outer
noise, appear as clearly distinguishable clusters
(concentrations of dots) in the 3PARD diagram
[2,3,4]. Each cluster from the 3PARD can be
selected individually and, by means of a back
transformation algorithm converted into a PRPD
pattern, thus making possible the identification of
separate PD sources. An example of the 3PARD
separation procedure is presented in Figure 3.
Three clusters from the 3PARD diagram are

analyzed and their back transformation to PRPD

pattern is performed. At the end the verification of
the identified type of the defect is performed with
an automated pattern classification application. It is
based on probabilistic (pattern recognition) and
deterministic
(knowledge-based
analysis)
approaches [3,4]. The automatically generated
report is presented in Figure 4.
In Figure 3, the external disturbances (cluster 1)
and background noise (cluster 2) are separated
from internal PD like phenomena. The pattern of
cluster 3 shows surface discharges in the end
winding area of the phase T. Patterns of similar
defect are identified by analyzing clusters 4 and 5.
Back transformation of clusters 6-8 has also been
performed. Automated pattern classification was
applied for the PRPD pattern corresponding to
cluster 6 and the report is presented in Figure 4.
According to the automatically generated report
this pattern shows delamination of the insulation
layers (class rmS2 table 1). Similar data
evaluation procedure was applied to PRPD
patterns of clusters 7 and 8. The findings reason
for the PD activity evidenced by all clusters - are
presented in Table1. The defects classes are
nominated as in [5].

CASE STUDY: TURBO GENERATOR 436

MVA, 20 KV

On-line PD monitoring is installed on twin 436

MVA, 20 kV turbo generators, manufactured in
2007. The architecture of the monitoring system is
shown in Figure 5. Installed PD sensors are 1.1nF
coupling capacitors.
Another key element in the separation of PD
sources from each other and from the background
noise is the digital band-pass filter of the
acquisition unit. Center frequency and bandwidth
can be freely chosen in the range from dc up to
30 MHz (with a selectable filter bandwidth from
9 kHz up to 3 MHz) to avoid disturbances and to
reach an optimal signal-to-noise ratio (SNR), even
under noisy on-site conditions. In this case the PD
measurements in the Acquisition Unit are set to a
center frequency of 1 MHz and bandwidth of 300
kHz in order to get an optimal SNR.

Figure 3: An example of automatic diagnosis of

the state of insulation

Figure 4: Automated pattern classification report

Table 1: Classification of PD defects
Defects
class

Defects type

rmS1

Discharges in micron cavities

rmS2
rmS3

Delamination of the insulation layers

Delamination of the insulation on the copper
side
Delamination of the insulation on the core side
End-winding surface discharge (tracking)
End-winding surface discharge (because of
insufficient space)
Discharge between corona protection and
stress grading layers

rmS4
rmE1
rmE2
rmE3

Figure 5: Architecture of the on-line PD monitoring

system
The start-up process of Generator 2 was monitored
and in Figure 6 the charge variation (green curve)
during the increase of the voltage (red curve) is
shown. The increase of PD activity can be
correlated to the changes in load conditions (zones
2 and 3). The PD value increased from a few pC
(background noise level) to 3.7 nC (Zone 2) and
then to 5.7 nC (Zone 3) when the machine was at
full load. The displayed voltage value is measured
at the input to the acquisition unit. To get the real
voltage value on the bus bar these values have to
be multiplied by a divider factor of about 1000.

a.

b.

Figure 6: PD trend during the start-up process.

The three phase PRPD pattern of zone 3 is
presented in Figure 7a. The 3PARD feature was
applied to identify PD defects in the PRPD
diagram. The back transformation of cluster 1 from
the 3PARD diagram was performed (Figure 7b)
and the equivalent PRPD pattern is presented in
Figure 7c. This pattern has similarities with those
generated by surface discharge in the end winding
area, with the highest amplitude in phase S. The
moisture contamination of the insulation surface
during the outage might be one of the reasons for
this type of PD. It should diminish with the
operation time of the machine. The PD activity
should remain under control. Clusters 2-4 from
3PARD diagram were discarded as they do not
originate from the insulation system of the
generator.

c.
Figure 7: PD data evaluation using 3PARD feature

CASE STUDY:
180MVA, 16.5KV

TURBO

GENERATOR

This case study shows an example where

preventive maintenance activity was triggered by
the results from on-line PD monitoring. The device
under monitoring is a 180 MVA turbo generator.
Figure 8 shows the three phase PD trend diagram
and the PRPD patterns and 3PARD diagrams
saved under each measured value.

Figure 8: Three phase PD trend and PRPD patterns before and after maintenance

According to the trend diagram, the increase of the

PD signal in phase W (blue trend) from 20 nC to 75
nC within two months (June August) was noticed.
By analyzing the PRPD patterns within this time
interval, the increase of PD activity generated by
surface discharge in the end winding area was
suspected. The separated PRPD pattern of this PD
source is presented in Figure 9.

Figure 10: Three phase PD trend diagram

Figure 9: PRPD pattern showing end winding

surface discharge separated using 3PARD
Consequently, the generator was taken out of
service for further investigations. The high
concentration of surface contaminants was
confirmed as one of the causes of the steep
increase of discharge activity. During the
maintenance activity the end section of the stator
windings was cleaned and no more surface
discharges were detected, see Figure 8 (after
maintenance section). Based on the PD results
obtained before and after the maintenance of the
machine, the effectiveness of the maintenance
program can be evaluated and the time when the
generator has to be taken out of service for
maintenance can be decided.
6

After the separation, the back transformation to

individual PRPD pattern was performed and
automated pattern classification was applied. The
results indicated advanced delamination of the
insulation layers of the stator bars, typical in
machines in service for more than 20 years.
PRPD patterns indicating the same type of PD
source were discovered in the other phases of the
machine. They show a lower PD level either
because the delamination stage is less advanced
or the PD sources are located deeper inside the
windings, therefore subjected to more attenuation.
Similar symptoms of insulation delamination were
also observed on the other identical generators in
the plant.

CASE STUDY: HYDRO GENERATOR 140

MVA, 16.5 KV

On line PD monitoring was installed on eight

identical hydro generators into operation since
1987. Preventive maintenance was triggered by
the high PD values recorded at one of these
machines (G1). By analysing the three phase PD
trend diagram of G1 (Figure 10) over three months,
the increase of the PD level in phase S was
observed (yellow trend). This indicated the
existence of a PD source in this phase.
Furthermore, data post-processing and automated
pattern classification were applied in order to find
the source of the high PD amplitude signal. The
PD level is measured using 1.1 nF coupling
capacitors and the digital filter was set to center
frequency of 3 MHz and frequency bandwidth of
300 kHz. The three phase PRPD pattern was
analyzed and the 3PARD tool was applied to
separate all PD sources from each other and from
the background noise. Figure 11 shows the
separation of the main PD source in phase S from
background noise.

Figure 11: Advanced post-processing of PD

CONCLUSIONS

Continuous PD monitoring is an essential

diagnostic tool for condition based maintenance to
guarantee high reliability of rotating machines.
The main key to performing appropriate diagnosis
of the state of the insulation is to accurately
separate different PD sources from external noise
and disturbances. To achieve this, synchronous
multi-channel evaluation techniques are applied.

[2] W. Koltunowicz, A. Obralic, A. Belkov and R.

Plath, "Continuous PD Monitoring of Stator
Insulation", in the proceedings of CMD 2010,
Tokyo, 2010.
[3] W. Koltunowicz and R. Plath, "Synchronous
multi-channel PD measurements", IEEE
Transactions on Dielectrics and Electrical
Insulation, vol.15, no.6, December 2008, pp.
1715-1723.

In order to identify PD defects, an advanced

automated system is proposed and successfully
applied.

[4] A. Belkov, W. Koltunowicz, A. Obralic, R. Plath

"Advanced Approach for Automatic PRPD
Pattern Recognition in Monitoring of HV
Assets", published in the proceedings of IEEE
ISEI, San Diego, 2010.

[5] IEC 60034-27 TS Ed.1.0, Rotating electrical

machines.

REFERENCES

[1] G. Stone, "Importance of bandwidth in PD

measurement in operating motors and
generators", IEEE Trans. on Dielectrics and
Electrical Insulation, Vol. 7, No.1, Feb. 2000.

INNOVATION FOR SECURE AND

EFFICIENT TRANSMISSION GRIDS
21, rue dArtois, F-75008 PARIS
http : //[Link]

CIGR Belgium Conference

Crowne-Plaza Le Palace
Brussels, Belgium | March 12 - 14, 2014

PD Testing and Monitoring of HV XLPE Cable Systems

W. Koltunowicz, M. Krger, L.V. Badicu, A. Piccolo

OMICRON Energy Solutions GmbH
Germany

SUMMARY
The paper deals with on-site PD testing, diagnostics and continuous monitoring of HV XLPE cable
systems and describes the method that can detect and locate partial discharges at all accessories
simultaneously. The examples of after-installation PD measurement procedure applied on 220 kV
cable line as well as in-service PD measurements are shown in the paper. For continuous monitoring
an advance concept to assess and monitor the condition of the insulation of a 420 kV cable system in a
10 km underground tunnel is described. Partial discharges are continuously monitored at all 400 kV
joints and terminations. At the same time, the system performs measurement of oil pressure in 400 kV
terminations as well as the condition of all sheath voltage limiters placed in the 400 kV joint bays.
Major attention is paid to the description of the technology of the partial discharge acquisition unit and
the inductive power supply solution. The advanced features for the elimination of disturbances and for
separation of different types of insulation defects based on synchronous, multi-channel and multifrequency techniques are shown. The processing of monitoring data is implemented in a modular
software system that allows reliable long-term storage of monitoring data and provides remote access
via a web interface. The separate chapters are dedicated to the procedure of site acceptance for the
monitoring system and also to system maintenance strategy.

KEYWORDS
HV XLPE Cables, after-installation test, partial discharges, continuous monitoring

[Link]@[Link]

INTRODUCTION

Cross-linked polyethylene (XLPE) with its excellent insulating properties is used both for new cable
installations as well as for replacing old cable systems up to 500 kV. The correct design of the XLPE
cable and its accessories (terminations and joins) is checked by the type tests and the quality of
production is checked by routine tests at the manufacturers plant according to relevant standards.
Installation work on site has an additional risk of introducing faults, as conditions are not ideal for an
installation of EHV accessories. Small particles, dust, moisture, etc. might lead to defects in
electrically critical locations of the accessories. Dielectric tests performed on site do not replace type
tests and routine tests. They are supplementary to dielectric routine tests and aim at checking the
dielectric integrity of the fully assembled cable line in order to eliminate defects such as damages
during transportation and lay-out, or incorrect assembly of the accessories. On-site tests are not
intended to check the design but they shall mainly detect any critical defects after installation on site,
which may decrease the withstand voltages of the new installation below the coordination withstand
voltages during on-site tests and later in service.
The most important stress situations for high-voltage cables during normal operation arise from the
operational alternating voltage and from switching over-voltages. Consequently, the preferred voltage
for on-site tests is ac voltage of industrial frequency. For on-site testing a larger frequency tolerance is
acceptable but frequencytuned circuits can easily be handled. Voltage testing delivers only binary
results (withstand or breakdown). Therefore, it is recommended to combine ac voltage testing with
sensitive on-site partial discharge (PD) measurements [1, 2]
A major part of all in-service failures in HV XLPE cables can be attributed to the insulation system of
accessories (joints and terminations) [2]. These failures will normally develop over time, due to the
ageing process during the service life or due to local degradation in the material. In order to detect
these changes at an early stage, detailed information on the actual insulation condition is necessary.
With suitable sensors this information can be derived by monitoring partial discharge activity during
the operation of the equipment.
This paper describes the best practice for performing after installation testing, checks in service, as
well as continuous monitoring of PD to assess the quality of the HV XLPE cable system and presents
a key tool to significantly reduce the risk of forced outages.
II

PD MEASUREMENTS DURING AFTER-INSTALLATION TESTING

HV tests are executed on site for all extruded HV cables and on-site test procedures have to be agreed
between the manufacturer and user on the basis of international and national standards.
Two IEC standards describe after installation tests of extruded cable systems: IEC 60840:2004 is
related to cables of rated voltages from 30 kV (Um = 36 kV) up to 150 kV (Um =170 kV) and
IEC 62067:2001 to cables of rated voltages from 150 kV up to 500 kV (Um = 550 kV) [3, 4]. The AC
test voltage level for the on-site test of new cable systems depends on the cable rated voltage and is
between 1.7 U0 and 2.0 U0 for rated voltages between 30 kV and 150 kV [2]. At higher rated voltages
the test voltage decreases from: 1.4 U0 (220-230 kV) to 1.3 U0 (275-345 kV), 1.2 U0 (380-500 kV) and
1.1 U0 for cables of 500 kV rated voltage.
High test power, especially required for long cable lines testing, can only be effeciently generated by
mobile resonant test systems, where the weight-to-power ratio and feeding power demand is relatively
low and the transport volume is acceptable (figure 1).

Fig.1.

Resonance test set connected to cable termination

The voltage during the dielectric test should be applied in steps (figure 2). PD measurements should be
taken at every voltage step. By increasing the test voltage in steps of e.g. 20% of the maximum test
voltage, critical defects are usually identified before breakdown.

Fig.2.

An example of voltage application in steps during dielectric on-site testing of 220 kV XLPE cable system

Because HV cables must be tested at the manufacturing facility prior to shipping to the installation
site, the on-site PD measurements focus on the field-installed accessories. For this purpose, each
accessory has to be equipped with special sensors to pick up the PD signal with high efficiency (figure
3). In one particular case of after installation testing of 220 kV XLPE cable line, 15 high frequency
current transformers (HFCTs) were used to pick up the PD signal [5]. They were equipped with an air
gap in the core to prevent magnetic saturation of the core.

Fig.3.

PD test arrangement

The PD system, HFCTs (MCTs100) connected to AUs (MPD600), was checked before installation
with calibration pulses injected to all the units connected in the daisy chain. On site, one MPD600 unit
was connected to the coupling capacitor and calibration was performed according to the IEC 60270
and IEC 60885 standards [1]. For the other 15 PD AUs mounted into the link boxes and close to the
terminations and connected in daisy chain with fibre optics (FO), the check of performance was
performed. A real calibration was not possible because impulses cannot be injected directly into the
closed and buried joint. The calibrator was connected to one turn of wire through the core of the
HFCT.
The resonance test set used IGBTs as switching semiconductors. They produce strong impulses,
which cause high interference impulses on the MPD600 instruments, especially to those units which
are rather close to the resonance test set. To eliminate this effect, PD signal gating was performed by
one extra MPD600 unit (with antenna) that was installed close to the IGBT circuit to receive the
switching impulses. The result of gating is shown in figure 4.

Fig.4. PD measurement at the coupling capacitor without gating (left) and with gating (right)

The final test voltage was 180kV, the test frequency was 25.95Hz and the test current was 68.1A. The
measurement frequency for all MPD600 units was set in the range from 2MHz to 3MHz. The
measurements at lower frequencies were not possible due to the strong interferences.
The PD results at 180kV of all MPD600 units are shown in figure 5. The tests could be finished within
3 days including the safety instructions, mounting of MPD600 to all link boxes and changing the
jumpers to all phases. All tests were successfully concluded and no partial discharges could be
detected above the noise level.

Fig.5. PD measurement at 180kV with 15 MPD units - 1.1=coupling capacitor, 1.3=joint 1-2, 1.4=joint 2-3,
1.5=joint 3-4, 1.6=joint 4-5, 1.7= joint 5-6, 1.8= joint 6-7, 1.9=joint 7-8, 1.10=joint 8-9, 1.11=joint 9-10,
1.12=joint 10-11, 1.13=joint 11-12, 1.14=joint 12-13, 1.15=end terminal GIS

III IN-SERVICE PD MEASUREMENTS AND TEMPORARY MONITORING

The PD tests were performed synchronously at GIS and OHL terminations in three phases of 220kV
XLPE power cable of 100 m length (figure 6). Six acquisition units, one portable PD analyzer
(OMS605) and three MPD600, connected with fibre optics, were used to perform synchronous
measurements at all terminations. Inductive HFCT sensors were used to couple the PD signal
(figure7). Their installation did not require the outage of the line.

Fig.6.

Test set up

Fig.7.

Installation of HFCTs and PD AUs

Multi-channel (3PARD) and multi-spectral (3CFRD) techniques were applied to analyse the results
and eliminate the background noise [6]. Figure 8a shows the synchronous PRPD patterns in three
phases of the GIS cable terminations and their equivalent 3PARD diagrams. Several clusters are
visible in the 3PARD diagram. The most relevant clusters were selected and the back transformation
to the PRPD patterns was performed. External disturbances have been successfully separated from
PD-like signal in phase S (figure 8b). As calibration of the set up was not possible in on-line
conditions, the PD values displayed in the PRPD patterns provide no significant information.

a)

b)
Fig.8. PRPD pattern before (a) and after separation and back transformation (b)

A monitoring session at the GIS side of the cable was initialized after the on-site PD measurements.
PD data was acquired for a centre frequency of 5 MHz and bandwidth of 1 MHz for 18 hours and had
been stored every 10 minutes. These values build up the PD trend diagrams (figure 9). In the figure,
the random appearing peaks of disturbances can be observed. The Besides the PD values, PRPD
patterns and 3PARD diagrams were stored in the database as pictures for each measuring point.

Fig.9.

PD trend in GIS termination of phase S

IV CONTINUOUS MOITORING SYSTEM: CONCEPT AND COMPONENTS

A continuous monitoring system is applied to the cable system in 10 km length tunnel. The tunnel
houses a new 400 kV, 2500 mm2 XLPE cable circuit. The longest cable sections are approximately
1176 m. PDs are continuously monitored at all joints and terminations. At the same time the system
performs measurement of oil pressure in terminations and check the condition of all sheath voltage
limiters (SVLs) located in the joint bays.
The concept of applied continuous monitoring system is presented in figure 10 [7, 8]. The signals from
different sensors measuring partial discharges, distributed temperature, oil pressure in terminations and
sheath voltage limiters are acquired by multi-channel data acquisition units.

Fig.10. Concept of monitoring system

In case of PD signals, the acquisition unit performs advanced pre-processing of the raw data. The
disturbances are removed and main characteristics of the PD signal are determined. The output of the
data pre-processing is transferred to a server that enables long-term data storage. Advanced intelligent
pre-processing reduces the amount of data to adequate levels for transmission over a communication
network.
HFCT sensors mounted on cross-bonding (CB) links are used to detect PD directly at the accessories
(figure 11). The PD acquisition system consists of one four-channel acquisition unit for each
accessory, connected to a data concentrator. One data concentrator collects monitoring data from two
or three acquisition units via optical fibres and routes them to a server. Pre-processing functions are
already performed in data concentrator like e.g. band passes integration, gating and denoising.
The active components of the monitoring system need electrical power for several processes. Beside the
computer and communication devices, also pre-amplifiers or signal converters (electrical to optical,
e.g.) located close to the sensors have to be supplied with power. The Inductive Power Supply provides
the necessary electronics to supervise and manage the dc current delivered at its output depending on
the various HV cable current load situations (figure 12). The electronics elements are filtered and
optimized to avoid any disturbances of the PD measurements close by.

Fig.11.

Schematic diagram of monitoring system with PD acquisition system (left)

SHEATH VOLTAGE LIMITERS MONITORING

Sheath Voltage Limiters minimize the transient voltage across the screen separation of cross-bonding
joints during switching or lightning transients and reduce the risk of damaging these. It is accepted
practice to install HV and EHV cable systems in tunnels with open air SVLs connected in delta
formation directly across the joint insulation flanges. A major advantage of such a bonding
configuration is the avoidance of grounding points within the tunnel normally required at each joint
bay. An additional benefit is that the voltage drop across the screen separation is smaller. The
disadvantage is that the rating of the SVLs must be higher and therefore the physical size bigger. The
access to the SVLs is limited, so there is a requirement to continuously monitor their status, such as:
x Normal operation (below inception voltage);
x SVL is short circuited when the conductive flashover trace is generated through the SVL varistors;
x SVL is in open loop totally damaged (active elements destroyed-exploded).
A SVL is a non-linear resistor and together with the parallel cable screen at a cross-bonding link
creates characteristic loop impedance. This impedance will change according to the status of the SVL
and the operation of SVL monitoring system is based on these changes. In the system, the PD
acquisition units located at the joint bay periodically inject high frequency signal pulses (using their
internal test generators). These signals propagate through the HFCT sensors to the SVL impedance
loop and finally arrive at the acquisition unit itself. The Fourier transforms of injected and returned
signals are calculated, averaged and compared. No special sensors are required for the scope and
major part of evaluation of SVL monitoring data is performed within the Monitoring Server.
Differences can be easily analysed and distinguished by spectral processing algorithm implemented in
the Monitoring Server Software.
SERVER AND SOFTWARE ARCHITECTURE
The server receives data for analysis, display, and storage. The acquisition units are configured and
remote-controlled by the Monitoring System Software. The Software supports remote access over
TCP/IP. This allows operators to quickly react to detected problems and access the stored data from any
remote location. The software is a highly modular, scalable distributed system. Its system architecture
consists of the windows-based core part and the web-based control part (figure 13). The core part of the
Monitoring Software is realized as windows services and runs continuously without any direct user
interactions. The core system implements: collection and persistence of measurement, data postprocessing and analysis, security tasks for data access and system operations, external interfaces for
data exchange over Ethernet or field bus.

Fig.12. Inductive Power Supply

Fig.13. Software architecture

TREND ANALYSIS
The monitoring system provides data from each of the PD and oil pressure sensors in two time
intervals: permanent and periodic one (Table I). During the permanent mode, the data are acquired
every 2-3 seconds, compared with threshold values and in real time displayed in graphical user
interface (figure 14). In case those data are within normal margins, they will be coloured in green. If
the values exceed thresholds for "warning" or "alarm" they are coloured in yellow or red accordingly.
Periodic measurements are initiated in equidistant time spans, e.g. every hour. Duration of the periodic
measurement is normally 1 min. During this time span all mentioned scalar values are calculated and
PRPD (phase resolved PD) and 3PARD diagrams are acquired. These data are saved for later postprocessing and trend visualization. Unscheduled periodic measurements are triggered in case of one or
more measured quantities exceeding the threshold level.

TABLE I. DEFAULT TIME SCHEDULING FOR DIFFERENT MEASURED VALUES

Mode
Value

Permanent

Periodic

Partial Discharge

every 2-3 sec

for 1 every 1 h

Oil pressure at terminations

every 2- 3 sec

once every 1 h

SVL status

--

once every 8 h

Fig.14. Graphic user interface

Fig.15. Installation of the system on site

IV ACCEPTANCE OF THE MONITORING SYSTEM ON SITE

The monitoring system was routine tested in the factory and later installed on-site (figure 15). The site
check of performance of the monitoring system was performed according to the following steps:
x Step 1: verification of the functional readiness of the measuring system and of the monitoring server
x Step 2: verification of the synchronous behaviour of the PD measuring system
x Step 3: determination of PD impulse attenuation, damping and dispersion along the cable system.
The following parameters were determined: PD impulse attenuation, damping and dispersion along the
cable, velocity of the calibration signal in the cable, best frequency ranges for PD measurements at all
PD units (with highest signal to noise ratio), PD detection path division factor for every chosen
frequency range.
V MAINTENANCE OF THE SYSTEM AND CUSTOMER SUPPORT
The services and maintenance scheme of the monitoring system is presented in Table II. The installed
system elements i.e. sensors, acquisition units, power supplies, batteries and fibre optical data
transmission network elements are periodically inspected and checked. These visual inspections and
functional checks include adjustments, repair or minor maintenance activities. Such visual inspections
are planned once every three years in coordination and align with the maintenance schedule activity of
the cable system. This would require access to the system installations, cable tunnel, manhole, shafts
and other related substation facilities including eventual outage of the cable system which requires
respective scheduling efforts. Software updates are foreseen. They would include modifications, bug
fixes, and potentially software enhancements with new features.
TABLE II. MAINTENANCE SCHEME OF MONITORIN SYSTEM
System
element
Hardare
Sofware

Maintenance scheme
Activity to be
performed
Visual check

Yearly

Functionality check

Every 3 years

Updates

Every 3 years

Owner
Owner &
system provider
System provider

Data evaluation

Periodic reports

System provider

Expert consultancy

In case of PD event

System provider

Periodicity

Responsibility

Regular checks and evaluation of data values and trending for the acquired partial discharges stored in
the database are also possible, if requested by the system owner. In case of repetitively reported
trending alerts or alarms or specific PD events detected by the monitoring system, PD Expert
consultancy and support can be requested by the system owner.
VI CONCLUSIONS
It can be recommended during after-installation testing to combine ac voltage testing with sensitive
on-site PD measurements;
PD measurements should focus on the field-installed accessories and the method should detect and
locate partial discharges at all accessories synchronously.
The monitoring system provider shall support the asset owner in all stages of the monitoring
project, from system design to maintenance of the monitoring system in service;
Separation of PD sources and suppress of external noise is performed by multi-channel evaluation
techniques;
The Inductive Power Supply provides the necessary power to the monitoring equipment when no
power supply is available;
To verify the status of SVL, the system utilizes the CB link loop impedance;
A modular, distributed monitoring software system allows reliable long term storage of monitoring
data and provides access via web interface.
REFERENCES

[1]
[2]
[3]

[4]

[5]
[6]
[7]

[8]

IEC 60885-3, Electrical test methods for electric cables Part 3: Test methods for partial
discharge measurements on lengths of extruded power cable
CIGRE Working Group 21-09 After Laying Tests on High Voltage Extruded Insulation Cable
Systems, ELECTRA No. 173 AUGUST 1997
IEC 60840, Edition 3.0 2004-04 Power cables with extruded insulation and their accessories
for rated voltages above 30 kV (Um= 36 kV) up to 150 kV (Um= 170 kV) Test methods and
requirements
IEC 62067, Edition 1.1 2006-03 Power cables with extruded insulation and their accessories
for rated voltages above 150 kV (Um= 170 kV) up to 500 kV (Um= 550 kV) Test methods
and requirements
M. Krger, R. Hummel, S. Bhler, M. Ufferhard, A. McGuigan, "Partial discharge measurement
and Monitoring on High Voltage XLPE Cables", CIGRE Symposium in Auckland, 2013
W. Koltunowicz and R. Plath, 2008, "Synchronous multi-channel PD measurements", IEEE
Transactions on Dielectrics and Electrical Insulation, vol.15, no.6, pp. 1715-1723, 2008
D. Gieselbrecht, W. Koltunowicz, A. Obralic, T, Ritz, P. Christensen, B. Schneider, K.H.
Cohnen, Monitoring of 420 kV XLPE Cable System in Underground Tunnel, published in
proceedings of the 2012 IEEE International Conference on Condition Monitoring and Diagnosis
(CMD), September 23-27, 2012, Bali, Indonesia
W. Koltunowicz, A. Obralic, A. Belkov, D. Gieselbrecht,"Continuous PD Monitoring of HV
XLPE Cable Lines", paper B1.9, in proceedings of: 8th International Conference on Insulated
Power Cables (Jicable), Paris-Versailles-France, 19-23 June, 2011.

Challenges in Monitoring of HV Equipment

W. Koltunowicz, L.V. Badicu, A. Piccolo
OMICRON Energy Solutions GmbH
Berlin, Germany
[Link]@[Link]

AbstractThe challenges in monitoring HV power cables,

transformers and rotating machines are presented in this paper.
In the case of monitoring HV power cables, the monitoring
system application to check the status of sheath voltage limiters is
described. Power supply solutions for active components of
monitoring cable systems in underground tunnels face different
constraints and the solution to the problem with the application
of inductive power supply is presented. For monitoring HV
transformers, the challenge to accurately measure the absolute
values of bushing insulation capacitance and tangent delta is
shown. The system also monitors partial discharge activity and
the sensitivity of the PD measurements in the UHF range is
discussed and compared with conventional PD measurements.
For the monitoring of rotating machines, examples of PD source
separation and discrimination from noise achieved with
advanced PD denoising and source separation using selectable
centre frequency, multi-channel and multi-spectral classification
systems are provided.
Keywordshigh
discharges;

voltage

I.

equipment;

monitoring;

partial

INTRODUCTION

The monitoring of HV equipment is a challenging task.

The customer requirements are different and guided by
applied maintenance strategy and varied service experience.
The monitoring systems described are provided for new HV
equipment to observe the state of the insulation in the
beginning and critical time of service life, as well as for older
equipment to determine the remaining lifetime. These systems
generally have versatile elements, e.g. sensors and acquisition
units, used in application with different assets but in most of
the cases, particular customer requests lead to the development
of new hardware components or software features.
For partial discharge (PD) monitoring systems on 420 kV
power cables laid in the underground tunnels, the challenge
was to provide an inductive power supply to PD acquisition
units located at the joints, since the power supply in the tunnel
is not permitted due to safety reasons. For the same type of
installation, the request to provide a simple, cost effective
system to check the status of sheath voltage limiters located at
cross-bonding joints also needs to be fulfilled [1].
For the monitoring of bushings insulation with high
accuracy, the absolute values of capacitance and dissipation
factor of the bushing insulation are requested. It was proposed
to use the reference voltage value from a nearby voltage
transformer. The system should also monitor PD activity by
the sensors at the bushing taps and inside the transformer tank

and utilize synchronous, multi-channel recording. The

sensitivity of the PD measurements in the UHF range, with an
antenna installed inside the oil drain valve is discussed and
compared with conventional PD measurements at measuring
bushing taps [2].
For the monitoring of PD in rotating machines, the effective
PD source separation and discrimination from noise is a must.
It can be achieved with an advanced PD denoising and source
separation using selectable center frequency, multi-channel
and multi-spectral classification system. A case study that
confirms the performance of the monitoring system is also
presented [3].
II.

MONITORING OF HV POWER CABLES

A. Status Check of Sheath Voltage Limiters

Sheath Voltage Limiters (SVL) minimize the transient
voltage across the screen separation of cross-bonding joints
during switching or lightning transients and reduce the risk of
damaging them. The access to the SVLs in the underground
tunnel is limited, so there is a requirement to continuously
monitor their status, such as:
x Normal operation (below inception voltage).
x

SVL is short circuited when the conductive flashover

trace is generated through the SVL varistors.

SVL is in open loop totally damaged (active

elements destroyed exploded).

A SVL is a non-linear resistor, and together with the

parallel cable screen at a cross-bonding link, creates
characteristic loop impedance. This impedance will change
according to the status of the SVL and the operation of SVL
monitoring system is based on these changes. In the system, the
PD acquisition units located at the joint bay periodically
inject signal pulses (using their internal test generators).
These signals propagate through the HFCT sensors to the SVL
impedance loop and finally arrive back to the acquisition unit
itself. The Fourier transform of injected and returned signals
are calculated, averaged and compared. No special sensors are
required for the scope and a major part of SVL monitoring
data evaluation is performed within the Monitoring Software.
Differences can be easily distinguished and analyzed by a
spectral processing algorithm called Correlation Coefficient.
It is based on a statistical comparison of the averaged spectra
between the reference data set and current SVL measurements.
The reference data set is built during the initial period of
system operation with SVL in normal operation condition. The

Correlation Coefficient Threshold (CCT) is calculated and

represents the maximum deviation of reference test impulses.
At the Monitoring Stage, the frequency spectra of all other
impulses measured during the normal monitoring stage is
compared with the CCT and if the deviation is bigger than
CCT, it means that the impedance and consequently the
condition status of SVL have changed. This system principle
has been preliminary tested (Fig. 1) and proven in the wooden
mock-up tunnel and finally successfully checked after cable
installation in an underground tunnel [1].

bushing taps are added to obtain the imbalance current. The

changes of the system voltage phase angle due to the daily load
variation lead to significant changes of the bushing capacitance
and dissipation factor, introducing a measuring error [4]. To
increase the sensitivity of the measurements, reference and
comparison methods have become more and more popular [5].
Both of these methods require a reference signal for calculation
of dissipation factor and capacitance. Depending on the source
of such signal, relative or absolute measurements can be
performed (Fig. 3).

Fig. 1. HFCT frequency signal response for different status of SVL.

B. Inductive power supply

The active components of the monitoring system need
electrical power for several processes. In addition to the
computer and communication devices the pre-amplifiers or
signal converters (e.g., electrical to optical) located close to the
sensors also have to be supplied with power. Due to safety
reasons in many existing cable tunnels, no auxiliary cables are
laid close to the power cables and cable joint locations are not
connected to the local AC power supply. The Inductive Power
Supply provides the necessary electronics to supervise and
manage the DC current delivered at its output depending on
the various HV cable current load situations (Fig. 2). The
PCBs and electronic elements are filtered and optimized to
avoid any disturbances of the PD measurements close by. The
DC current delivered is depending on the cable load situations,
but even at very low load the power is sufficient to run the
monitoring system [1].

Fig. 3. Architecture of the monitoring system applied for comparison and

reference measurement methods.

With relative measurements, the bushing-to-bushing

comparison or dual transformer comparison is performed,
while with the absolute measurement the reference signal is
taken from voltage transformers (VTs) of the substation. Grid
unbalance, like phase shifts or changes in the system voltages,
can have a misleading impact on the relative measurements of
capacitance and dissipation factor. The simultaneous ageing of
the reference and test bushings cannot be detected. To reduce
the impact of these uncertainties, absolute measurements are
recommended. The reference signal is taken from a group of
VTs located near the monitored transformer. This pure resistive
signal is compared with the mainly capacitive leakage current
measured at the bushing tap of the transformer. The
temperature correction of the dissipation factor values is
performed and takes into account the insulation type of the
bushing [2].
In setting up the monitoring system, one capacitive tap
sensor is installed at each bushing to measure PD, capacitance,
dissipation/power factor and transient overvoltages (Fig. 4).
The system is equipped with multiple redundant protections to
guarantee safe operation. The ambient temperature and relative
humidity are also measured within the same tap sensor to
compensate the measured values.

Fig. 2. Inductive power supply.

III.

MONITORING OF POWER TRANSFORMERS

A. Accurate measurement of capacitance and dissipation

factor
Traditionally, the sum-of-currents method is the most
common method used in the monitoring of bushing
capacitance. The leakage currents measured at the three

The PRPD patterns are complex with several PD sources

overlapped. In order to separate clusters of different PD
sources, a synchronous multi-channel PD evaluation technique
(3PARD) is applied. The 3PARD diagram visualizes the
relationship between amplitudes of a single PD pulse in one
phase and its crosstalk generated signals in the other two
phases. By repetition of this procedure for a large number of
PD pulses, PD sources within the test object as well as outer
noise appear as a clearly distinguishable concentration of dots
in a 3PARD diagram. By examining individual clusters in the
3PARD diagram, a separation between noise and PD
phenomena is possible [6].

Fig. 4. Architecture of the monitoring system.

The diagnosis of the bushing is performed by analyzing the

trend of dissipation factor, its magnitude and rate of change.
The diagnosis of bushing capacitance is performed by
comparing the capacitance values from on-line monitoring
with the values from off-line measurements. The difference
between them gives the capacitance variation C that has to
remain within certain limits [2]. The accuracy of the
monitoring system is < 2 pF for capacitance measurement
and 0.01 % for dissipation factor measurement. The class of
VTs and its dynamic error has to be additionally taken into
consideration.
B. Combination of conventional and unconventional PD
detection
Two methods are recommended for PD measurements: the
conventional method (according to IEC 60270) with sensors at
the bushing taps and an unconventional ultra-high frequency
(UHF) method with a sensor placed inside the transformer tank
(Fig. 5). With the conventional method, the PD signal from
each tap is synchronously acquired by a three-channel
acquisition unit. The central frequency of the digital band pass
filter of the acquisition unit is selected to reach the optimal
signal-to-noise ratio. To obtain more detailed information
about the type and location of PD defects in the insulation,
unconventional UHF PD measurements are performed in the
frequency range between 100 MHz and 2 GHz with an antenna
installed inside the oil drain valve. The presence of external
noise in this frequency range is low and radio or mobile phone
signals are easily recognized and eliminated from the
measurements. PD activity inside the bushing insulation and
close to the end winding area is mostly detected with the
conventional method while the rest of the tank is covered by
the UHF antenna. Even if the PD signal is measured in a
different frequency range, the obtained PRPD diagrams have
very similar patterns that makes recognition of the defect type
easier. The signal detected by the UHF antenna is synchronized
with the signal detected at the bushing taps. Furthermore, the
measured pulses in the UHF range, mostly coming from
internal PD activity, can trigger the start of conventional
measurements [2].
An example of a three-phase PD trend of conventional
measurements is presented in Fig. 6. At each point in the trend
diagram,PRPD patterns and 3PARD are available.

Fig. 7 shows the back transformation to the PRPD pattern

of the clusters 1 and 2 from the 3PARD diagram. The patterns
of the clusters 1 and 3 appear to be generated by bubbles and
surface discharge with the highest amplitude in phase B
(cluster 1) and phase A (cluster 3). The shape and phase
position of the patterns of the clusters 2 (phase A) and 6 (phase
C) may indicate partial discharge activity inside the voids of
the insulation system. The clusters 4 and 5 are generated by
external interferences.

Fig. 5. Location of sensors for conventional (left) and UHF measurements.

Fig. 6. The 3-phase PD trend for conventional measurements with PRPD and
3PARD diagrams (right).

Internal PD activity was also detected with the

unconventional UHF method and detected in the frequency
range from 450 to 650 MHz (Fig.8 - left). The PRPD pattern
corresponding to a central frequency of 600 MHz is presented
in Fig. 8 (right).
The signal detected during the on-line monitoring was
synchronized with a 50 Hz signal taken from the measuring
tap of phase A. The voltage phase position of PD pulses
indicates the internal discharges located close to phase A. The

PRPD patterns of frequencies between 1 GHz and 1.4 GHz

were checked and no internal PD activity was found.

Fig. 7

3PARD diagram and PRPD patterns of the selected clusters.

Fig. 9
Fig. 8

IV.

An example of the 3PARD separation procedure.

Frequency sweep diagram (left) and PRPD pattern (right).

MONITORING OF ROTATING MACHINES CASE STUDY

The 1160 MVA, 27 kV turbo generator was put in

operation in 1998 and in 2012 was taken out of service for
regular maintenance. The on-line PD monitoring system was
installed during the outage and it started immediately collecting
data when the generator was put back into operation. After the
conditioning time of two months data processing was
performed with synchronous multi-channel evaluation
techniques [6] and an automated pattern classification system.
An example of the 3PARD separation procedure is presented in
Fig. 9. Three clusters from the 3PARD diagram are analyzed
and their back transformation to PRPD pattern is performed
[3]. In Fig. 9, the external disturbances (cluster 1) and
background noise (cluster 2) are separated from internal PD
like phenomena. The pattern of cluster 3 shows surface
discharges in the end winding area of the phase T. Patterns of
similar defect are identified by analyzing clusters 4 and 5. Back
transformation of clusters 6-8 has also been performed.
Automated pattern classification was applied for the PRPD
pattern corresponding to cluster 6 and indicated delamination
of the insulation layers (class rmS2 in TABLE 1) [7]. A similar
procedure was applied to PRPD patterns of clusters 7 and 8.
The findings in all clusters indicate the reason for the PD
activity and are presented in TABLE 1. The defect classes are
nominated as shown [8].

TABLE I.
Cluster
No.
6
7, 8
3, 4, 5

Defect
class
rmsS1
rmS2
rmS3
rmS4
rmE1
rmE2
rmE3

CLASSIFICATION OF PD DEFECTS
Defect type
Discharge in micron cavities
Delamination of the insulation layers
Delamination of the insulation on the copper side
Delamination of the insulation on the core side
End-winding surface discharge (tracking)
End-winding surface discharge (gap discharge)
Discharge between corona protection and stress
grading layers

V. CONCLUSIONS
The following monitoring system solutions, hardware
components and software features were successfully applied to
monitor HV equipment:
The Inductive Power Supply delivers the electrical
power to the active components of HV cable
monitoring system in tunnels.
The Monitoring Software evaluates the status of SVL;
no specific hardware elements were necessary.
The Multi-channel and multi-spectral classification is
used to separate PD sources and discriminates them
from noise.
The reference voltage signal taken from voltage
transformers of the substation allows the dissipation
factor and capacitance of the bushings to be calculated
with high accuracy.
The combination of conventional and unconventional
UHF PD measurements increases the sensitivity and
reliability of the measurements.

REFERENCES
[1]

[2]

[3]
[4]

D. Gieselbrecht, W. Koltunowicz, A. Obralic, T, Ritz, P. Christensen, B.

Schneider, K.H. Cohnen, Monitoring of 420 kV XLPE Cable System in
Underground Tunnel, published in proceedings of the 2012 IEEE
International Conference on Condition Monitoring and Diagnosis
(CMD), September 23-27, 2012, Bali, Indonesia.
L.V. Badicu, W. Koltunowicz, A. Piccolo, A. McGuigan, C. Feely,
"Monitoring of a Distribution Transformer at Winchelsea Substation", in
proceedings of TechCon 2014, Sydney, Australia.
L.V. Badicu, W. Koltunowicz, M. Koch and A. Piccolo, Return of
Experience from Continuous PD Monitoring of Rotating Machines, ISH
2013, Seoul, Korea.
Z. Berler, V. Prykhodko, J. Watson, J.S. Skinner, D. Bates, Analyzing
Data from On-line Continuous Bushing Monitoring Systems Paper 9,
proceedings of the International Conference on Condition Monitoring,

[5]

[6]

[7]

[8]

Diagnosis and Maintenance CMDM2013, Bucharest, Romania, pp.7378, 2013.

M. Koch, M. Krger and W. Koltunowicz, A New Method for On-line
Monitoring of Bushings, Transients and PD of Power Transformers,
Proceedings of The 4th International Advanced Research Workshop on
transformers, Baiona, Spain, pp. 287-290, 2013.
W. Koltunowicz and R. Plath, "Synchronous multi-channel PD
measurements", IEEE Transactions on Dielectrics and Electrical
Insulation, vol.15, no.6, December 2008, pp. 1715-1723.
W. Koltunowicz, A. Obralic, A. Belkov and R. Plath, "Continuous PD
monitoring of stator insulation", in the proceedings of International
Conference on Condition Monitoring and Diagnosis (CMD), 2010,
Tokyo, 2010.
IEC 60034-27 TS Ed.1.0, Rotating electrical machines.

Monitoring of a Distribution
Transformer at Winchelsea Substation
by
L.V. Badicu, W. Koltunowicz, A. Piccolo,
A. McGuigan, C. Feely

PROCEEDINGS OF TECHCON ASIA PACIFIC 2014

MONITORING OF A DISTRIBUTION TRANSFORMER

AT WINCHELSEA SUBSTATION
L.V. Badicu1, W. Koltunowicz1, A. Piccolo1, A. McGuigan2, C. Feely3
1
OMICRON Energy Solutions GmbH, Germany
2
OMICRON Electronics, Australia
3
POWERCOR Australia Ltd., Australia

Abstract
             

            
         

          
             


           

            
          
          
         
            
           


Background
The transformer is a 5/7 MVA ONAN/ONAF 66000/22000 V Dyn1 unit constructed in 1998. It
was initially put into service in a city environment in Melbourne where it operated
predominantly on a fixed tap and without any problems until removed from service in 2007
as part of a major upgrade project. Later in 2007, the transformer was relocated to
Winchelsea Zone substation (figure 1), a rural environment where it was placed in service to
replace a failed transformer. Once again operation was on a fixed tap until 2009 when the
tap changer was put into automatic operation and used to provide voltage regulation for
the station. The unit was not heavily loaded and gave no indication of any problems until
the annual oil sample dissolved gas analysis (DGA) result from February 2010, which
indicated 5400 ppm of hydrogen. The dissolved gas ratios indicated that partial discharge
(PD) was taking place.
In order to better monitor what was occurring, an on-line DGA monitor was installed giving
4 hourly readings. Initially the monitor recorded less H 2 ppm, but it was quickly realized that

this was actually a false reading due to the fact that the detection limit for hydrogen of 3000
ppm had been exceeded and, counter-intuitively, this was causing the monitor to read low.
The transformer was then partially de-gassed. The results for the next three years are
shown in figure 2. As can be seen, the hydrogen and methane levels showed a steady
increase, which seemed to indicate that PD was ongoing.

Figure 1
Monitored transformer at Winchelsea Zone substation

Figure 2
On-line DGA results

There have been several attempts to measure PD with limited success up until the
installation of the continuous OMICRON monitoring system, which has confirmed the PD
activity and facilitated some further analysis.

The installed monitoring system combines accurate measurement of the bushing

capacitance and dissipation/power factor with grid transient recording and PD
measurements (figure 3). It provides remote assessment of the dielectric integrity of
transformer and bushings with high accuracy comparable to off-line testing. The following
parameters are continuously monitored:
x Capacitance and dissipation/power factor of high voltage (HV) bushings;
x PD level at the bushings and inside the tank;
x Transient over-voltages at the HV bushings.

Figure 3
Parameters continuously monitored
Project-specific system design
Accurate measurement of capacitance and dissipation/power factor
Traditionally, the sum-of-currents method is the most common method used in monitoring
of bushing capacitance. The leakage currents measured at the three bushing taps are added
to obtain the imbalance current. The changes of the system voltage phase angle due to the
daily load variation lead to significant changes of the bushing capacitance and dissipation
factor, introducing a measuring error 1. To increase the sensitivity of the measurements,
reference and comparison methods have become more and more popular 2. Both these
methods require a reference signal for calculation of dissipation factor and capacitance.
Depending on the source of the reference signal, relative or absolute measurements can be
performed (figure 4) 3. With relative measurements, the bushing-to-bushing comparison or
dual transformer comparison is performed, while with absolute measurement reference
signal is taken from voltage transformers (VTs) of the substation. Grid unbalance, like phase
shifts or changes in the system voltages can have a misleading impact on the relative
measurements of capacitance and dissipation factor. The simultaneous ageing of the
reference and test bushing cannot be detected. To reduce the impact of these uncertainties
(figure 5), absolute measurements are recommended at the Winchelsea substation. The
reference signal is taken from the LV side of a group of VTs located near the monitored
transformer. This pure resistive signal (UVT) is compared with the mainly capacitive leakage
current (IB) measured at the bushing tap of the transformer. The corresponding vectors of
these two signals are rotating in the same four-quadrant coordinate system (figure 6) and

&
the capacitive current I B always leads in revolution. The angles between the reference axis
and the two vectors are permanently measured and, by computing the difference between
them, the angle and the power factor cos are obtained. The dissipation angle () and
dissipation factor (tg) can be calculated accordingly. The temperature correction of the
dissipation factor values is performed 4. It takes into account the OIP (oil impregnated
paper) type of the bushing. The formula (1) is used to calculate the bushing capacitance CB.
The IB current is measured by the monitoring system, while the U High value is directly taken
from VTs.

Figure 4
Architecture of the monitoring system applied for
comparison and reference method

Figure 5
Sensitivity/uncertainty
introduced by each method

Figure 6
Vector representation of the UVT and IB in a four-quadrant coordinate system
IB

2 S f C B U High C B

IB
2 S f U High

(1)

The accuracy of the monitoring system is < 0.8 pF for capacitance measurement and 0.01
% for dissipation factor measurement. The 0.5 class of VTs also has to be taken into
consideration (Table 1).

Table 1
Voltage error and phase displacement for measuring voltage transformers 5

Combination of conventional and unconventional PD detection

Two methods are recommended for PD measurements: the conventional method
(according to IEC 60270) with sensors at the bushing taps and an unconventional ultra-high
frequency (UHF) method with a sensor placed inside the transformer tank. With the
conventional method, the PD signal from each tap is synchronously acquired by a three
channel acquisition unit. The central frequency of the digital band pass filter of the
acquisition unit is selected to reach the optimal signal-to-noise ratio. To obtain more
detailed information about the type and location of the insulation PD defects, the
unconventional UHF PD measurements in the frequency range between 100 MHz and 2 GHz
with an antenna installed inside the oil drain valve are performed. The presence of external
noise in this frequency range is low and radio or mobile phone signals are easily recognized
and eliminated from the measurements. PD activity inside the bushings insulation and close
to the end winding area is mostly detected with the conventional method while the rest of
the tank is covered by the UHF antenna. Even if the PD signal is measured in a different
frequency range, the obtained PRPD diagrams have very similar patterns that makes
recognition of the defects type easier (figure 7). The signal detected by the UHF antenna is
synchronized with the signal detected at the bushing taps. Furthermore, the measured
pulses in the UHF range, mostly coming from internal PD activity, can trigger the start of
conventional measurements. Thus a better separation between internal and external PD
pulses can be obtained.

Figure 7
Example of PRPD patterns measured by conventional method (left) and UHF method (right)
for a known internal PD defect.

Measurements of transient over-voltages

The voltage transients appearing at the bushings and propagating into the windings create
thermo-electrical stress and accelerate the aging of the insulation. The number of the
transients and their shape (rise time, magnitude and duration) are important parameters.
The magnitude of the lightning-induced transients is reduced below 2 p.u. (phase-to-ground
voltage = 1 p.u.) by surge protective devices 6,7, but the fast transients coming as a result of
operations in the substation can also be dangerous 8. The magnitude and shape of the
transients strongly depends on the power grid configuration, length of the lines, type of
connecting HV equipment 6. Figure 8 shows the waveform of the transients when a cable,
gas insulated line (GIL) or overhead line (OHTL), all with the same length of 50 m, are
connected at the transformer. In figure 9, the variation of transient peak magnitude as a
function of the length of connecting terminals are presented. At the Winchelsea substation,
the OHTL of 30 km and 40 km are connected to the monitored transformer. As a result,
transients of high amplitude are expected to appear at the bushings of the transformer.

Figure 8
Waveform of the transients when a cable or
GIL or OHTL is connected at the transformer 7

Figure 9
Variation of peak magnitude of transients
as a function of the length of terminals 7

User-friendly visualization of monitoring data

The monitoring software runs on a central computer. It can either be directly accessed onsite or remotely using a trusted network via the convenient web-based interface. The
software is a management tool for system configuration, continuous data acquisition and
visualizing real-time or stored measurement data. There are two modes of collecting
monitoring data: permanent mode and periodic mode. During the permanent mode, the
data is acquired every second, compared with threshold values and displayed in real time.
When this data is within normal margins, the values are displayed in green (figure 9) and are
not stored in the database repository. In the periodic mode, data is measured at specific,
defined intervals or at the occurrence of a violation of the predefined threshold values. In
this mode, all measuring data is stored in the database.

Figure 9.
Real-time data displayed in the graphical user interface
System installation
The architecture of the monitoring system is shown in figure 10. For each bushing, one
capacitive tap sensor to measure PD, capacitance and dissipation/power factor is installed
(figure 11). It is equipped with multiple redundant protections to guarantee safe operation.
The ambient temperature and relative humidity are also measured within the same tap
sensor to compensate the measured values. The UHF antenna is put inside the oil drain
valve located in the upper part of the tank (figure 12) and is connected to UHF down
converter and finally to the fourth channel of the acquisition unit. The outdoor monitoring
server and communication modem are mounted on the tank of the transformer (figure 13).
Only two VTs (phase B and phase C) were available for reference. The calculation of the
dissipation factor for phases B and C is based on the reference signal taken from the VTs
installed on these phases. To calculate the dissipation factor of the bushing installed on
phase A, the reference signal is taken from both VTs. The resulting reference voltage vector
follows the changes of the other two phases in a dynamic way. During the commissioning,
calibration of the monitoring system was performed and the level of alarms and measuring
intervals was selected in the monitoring software. The threshold values for dissipation
factor and capacitance were chosen based on the recommendations in the standards and
experience:

Dissipation factor

The diagnosis of the bushing is performed by analyzing the trend of tg, the magnitude and
its rate of change. When tg reaches a predefined level of alarm 0.7% (table 2) 9
continuous operation is no longer recommended. On the other hand, bushings with values
of tg above a predefined level but with stable trend may stay in operation. When tg
doubles the value over six months, off-line investigations and additional measurements are
recommended.

Figure 10.
Architecture of the installed monitoring system

Figure 11.
Tap sensor for C, tg, transients
and PD measurements

Figure 12.
UHF antenna installed in oil
drain valve

Figure 13.
Server, acquisition units
and UPS

Capacitance

The diagnosis is performed by comparing the capacitance values from on-line monitoring
with the values from off-line measurements. The difference between them gives the
capacitance variation C that has to remain within certain limits. The change of bushing
capacitance when one insulation layer is short-circuited for bushings of different rated
voltage is presented in table 3 10. As the detailed information about the bushing design
(number of insulation layers) is not available, the limits of C variation was estimated to be
a maximum 10%, taking into account the type of the bushing and its rated voltage. Such a
defect does not necessarily lead to the total failure of the bushing, but it creates a higher
electric field on the healthy layers.

Table 2
Acceptance level of dielectric losses for bushings of
9
different design (at 20C)
Standards
DF (tan) IEC60137
PF IEEE C57.19.01

RIP
<0.7%

OIP
<0.7%

RBP
<1.5%

<0.85%

< 0.5%

<2%

Table 3
Voltage class and change of capacitance for
10
condenser type bushings
Voltage in kV
123
245
420

No. of layers
14
30
40

Change in %
7.1
3.3
2.5

PD detection with the conventional method requires a calibration of the measuring system
according to IEC60270. To fulfill this requirement, PD pulses of a known charge are injected
in the current path of the bushing and measured at their taps. The threshold for alarm level
has to be related to the routine acceptance test level, 500 pC at 1.5Um according to
IEC60067-3 and IEEE C.57.12.90. Different types of the defect generate signals of different
apparent charge magnitude and PD pulse repetition rate. The decision about the criticality
of measured PD signals requires analyses of PD patterns. Wide bandwidth or narrow
bandwidth UHF measurements can be applied. In wide bandwidth measurements, the
highest magnitude of the UHF signal is measured in the frequency range from 100 to 2000
MHz. In this case the noise can influence the result of measurements. To overcome this,
narrow band measurements are performed in the selected frequency range. With the
transformer out of service, the spectra of the UHF signal are acquired and the external noise
is identified. The central frequency and the bandwidth of the measurements are selected in
the "clean" area of the spectra. The spectra are verified when the transformer is back in
service and the frequency range of the measurements can be adjusted if necessary.
The amplitude of the transient over-voltages is calculated and expressed as a function of the
phase-to-ground voltage and is displayed in p.u.
Periodic measurements are initiated in equidistant, one-hour time spans. The duration of
the periodic measurement is normally 1 min. During this time, all scalar values (i.e.
dissipation factor, capacitance, charge, repetition rate of the PD pulses, etc.) as well as
images (i.e PRPD - phase resolved partial discharge patterns, 3PARD 3 phase amplitude
relation diagram 11, magnitude and shape of transient over voltages, etc.) are stored in the
database.

Analysis of the results

Capacitance and dissipation factor
The example of the trend of capacitance variation (C) is presented in figure 14. The
maximum value of C measured over the period of four months is compared with the
threshold value (figure 15). The C in phase B is the highest but still below the maximum 10
% limit of change.
The three-phase trend of the dissipation factor is shown in figure 16. All the values are
below the threshold values predefined for OIP bushings 9 and their trend is stable. A
comparison between the highest on-line measured value for each phase and the predefined
threshold is shown in figure 17.

Figure 14
Three phase capacitance and ambient
humidity variation (red trend)

Figure 16
Three phase dissipation factor variation

Figure 15
Comparison between the highest capacitance
variation (green) and the threshold limit (red)

Figure 17
Comparison of the highest measured value
of dissipation factor with the threshold limit

PD monitoring
PD monitoring with the conventional method
The three-phase PD trend is presented in figure 18. At each point visible in the trend
diagram, PRPD patterns and 3PARD are available.
Figure 19 depicts the PRPD patterns of the PD signal acquired by the three-channel
synchronous system. They are complex patterns with several PD sources overlapped. In
order to separate clusters of different PD sources, a synchronous multi-channel PD
evaluation technique is applied 11. The 3PARD diagram (figure 20) visualizes the relationship
between amplitudes of a single PD pulse in one phase and its crosstalk generated signals in
the other two phases. By repetition of this procedure for a large number of PD pulses, PD
sources within the test object as well as outer noise appear as a clearly distinguishable
concentration of dots in a 3PARD diagram 11. By examining individual clusters in the 3PARD
diagram, a separation between noise and PD phenomena is possible.
Figure 21 shows the back transformation to PRPD pattern of the clusters 1 and 2. The
patterns of the clusters 1 and 3 appear to be generated by bubbles and surface discharge
with the highest amplitude in phase B (cluster 1) and phase A (cluster 3). Similar PRPD
patterns were reported in 12. The shape and phase position of the patterns of the clusters 2
(phase A) and 6 (phase C) may indicate partial discharge activity inside the voids of the
insulation system 12. The clusters 4 and 5 are generated by external interferences

Figure 18
Three phase PD trend

Figure 19
Three phase synchronous PRPD patterns
PD monitoring with unconventional UHF method
Performing a frequency sweep, two spectra of the signal are obtained (figure 22). The upper
spectrum is built based on the maximum amplitudes of the time domain signal acquired at
each value of the frequency during the sweep. The lower spectrum corresponds to their
minimum amplitudes. Internal PD activity is always visible on the upper spectrum while
external interferences (corona discharge, radio waves, GSM) are visible on both spectra.
Phase C

Figure 20
Equivalent 3PARD diagram

Phase B

Phase A

Figure 21
Individual PRPD patterns of the selected clusters

Internal PD activity was identified in the frequency range from 450 to 650 MHz. The PRPD
pattern corresponding to a central frequency of 600 MHz is presented in figure 23. The
signal detected during the on-line monitoring was synchronized with a 50 Hz signal taken
from the measuring tap of the phase A. It can be seen in figure 23 that the phase of the
voltage where the PDs occur is the one characteristic to internal discharges. Furthermore, it
indicates a possible location of the PD activity, namely, in the vicinity of phase A. The PRPD
patterns of frequencies between 1 GHz and 1.4 GHz were checked and no internal PD
activity was found.

Figure 22
Frequency sweep diagram

Figure 23
PRPD pattern

Transients
Thirty transient over-voltages with the amplitudes between 1.25 and 1.77 p.u. were
recorded in four months after installation of the system (figure 24).
The transient with the highest amplitude (1.77 p.u.) is shown in figure 24a. Figure 24b and c
show other transient over-voltages recorded during the monitoring period. Even with such
amplitudes, these transients represent a threat to the insulation of the bushings and
windings because of the resonant phenomena which can lead to higher voltage distribution
between the turns. The number of transients cannot be controlled but their monitoring can
help identify defective equipment which generates over-voltages, near the transformer.

Figure 24
Time stamp and shape of the transients detected at the bushings

Discussion
As the dissolved gas ratios in the DGA system indicated that PD is taking place, the
OMICRON on-line monitoring system for PD, bushing capacitance and tg was installed on
the distribution transformer of POWERCOR. The PD activity was confirmed with both
conventional and unconventional UHF measurements. The synchronous multi-channel
evaluation technique separated clusters of different PD sources. Furthermore, it indicated a
possible location of the PD source in the vicinity of phase A.
The bushing capacitance and tg values remain within acceptable levels indicated in the
literature.
Concerning DGA analysis, high concentrations of H2 and CH4 indicate oil decomposition
under PD activity 13. The presence of high quantities of CH4 and C2H6 indicate an extreme
overheating of the mineral oil and adjacent metals. A low concentration of CO suggests that
PD activity may not take place in the vicinity of the paper insulated parts.
It is intended to carry out internal inspections following the installation of another
transformer at the site planned for 2014.
Acknowledgement
OMICRON sincerely appreciates the high level of support and co-operation of POWERCOR in
enabling this trial to proceed and with the installation of the monitoring system.
References
1. Z. Berler, V. Prykhodko, J. Watson, J.S. Skinner, D. Bates, Analyzing Data from On-line
Continuous Bushing Monitoring Systems Paper 9, proceedings of the International
Conference on Condition Monitoring, Diagnosis and Maintenance CMDM2013,
Bucharest, Romania, pp.73-78, 2013.
2. M. Koch, M. Krger and W. Koltunowicz,A New Method for On-line Monitoring of
Bushings, Transients and PD of Power Transformers, Proceedings of The 4th International
Advanced Research Workshop on transformers, Baiona, Spain, pp. 287-290, 2013
3. ***MONTRANO Continuous Monitoring System for Power Transformers Technical
Brochure, OMICRON Energy Solutions, Berlin GmbH, 2013.
4. Transformer Bushings, type GOB Installation and Maintenance Guide, ABB Technical
Brochure,[Link]
5c004a7dc9&db=db0003db004283
5. ***IEC60044-2:2003 Instrument transformers Part2: Inductive voltage transformers
6. X. Dong, S. Rosado, Y. Liu, N.C. Wang, E.L. Line and T.Y. Guo, Study of abnormal electrical
phenomena effects on GSU Transformers, IEEE Transactions on Power Delivery, vol. 18,
No. 3, July 2003.

7. A. Said, E.A. Badran, M.A. Abd-Allah, Mitigation of very fast transient overvoltages at the
more sensitive points of gas insulated substation, International Journal on Electrical
Engineering and Informatics, Vol. 4, No. 3, October 2012.
8. U. Riechert, H. Ito, E. Zaima, K. Uehara and W. Chen, Insulation Co-ordination for Very
Fast Transients in Gas-Insulated UHV Substations, presented at UHV Colloquium New
Delhi 2013, Session 2.3 Substations.
9. CIGRE TB 445 Guide for Transformer Maintenance CIGRE Working Group A2.34,
February 2011.
10. T. Stirl, R. Skrzypek, S. Tenbohlen, R. Vilaithong, Online Condition Monitoring and
Diagnosis for power Transformers their Bushings, Tap Changer and Insulation System
Paper? Proceedings of the International Conference on Condition Monitoring and
Diagnosis CMD 2006, Changwon, South Korea , April 2006.
11. W. Koltunowicz and R. Plath, Synchronous Multi-Channel PD Measurements IEEE
Transactions on Dielectrics and Electrical Insulation, Vol.15, No. 6, p. 1715-1723, 2008.
12. A. Carlson, J. Fuhr, G. Schemel, F. Wegscheider, Testing of Power Transformers
Routine tests, Type tests and Special tests, 1 st Edition, published by Pro Print, Zrich,
Switzerland, 2003.
13. ***IEC60599:1999, Mineral oil-impregnated electrical equipment in service Guide for
interpretation of dissolved and free gases analysis.
Biography
Laurentiu Viorel Badicu is HV Application Engineer at OMICRON Energy Solutions GmbH,
Berlin, Germany since 2012. He is responsible for maintenance of the installed on-line
monitoring systems, customer trainings, performing site measurements (partial discharge,
dissipation/power factor, capacitance etc.) and data evaluation. His interest covers also
aspects related to the ageing mechanism and condition assessment of the insulating
materials.
He received the Dipl.-Ing. and the Ph.D degrees in electrical engineering from University
Politehnica of Bucharest, Romania in 2008 and 2012, respectively.
Wojciech Koltunowicz is with OMICRON Energy Solutions GmbH, Berlin, Germany, where he
is involved in monitoring of HV equipment. From 1987 to 2007 he was with CESI, Italy,
where he was mainly involved in HV testing and diagnostics of HV equipment. From 1984 to
1987, he was a research scientist in the High Voltage Department at the Institute of Power
in Poland.
He received [Link]., PhD and [Link]. degrees in electrical engineering from the Warsaw
University of Technology in 1980, 1985 and 2004, respectively.

He is secretary of CIGRE Advisory Group D1.03 Insulating Gases, WG D1.25 and D1.37. He
is member of CIGRE AG D1.02 High Voltage and High Current Test and Measuring
Technique and Diagnostic and WG D1.51. He is also member of IEC TC42 WG14. He is
author of dozens of international reports.
Andrea Piccolo is HV Application Engineer at OMICRON Energy Solutions GmbH, Berlin,
Germany since 2013. From 2009 to 2011 he was with TechImp SpA as R&D engineer. From
2011 to 2013 he was responsible for electrical diagnostics within Nidec ASI (former Ansaldo
Sistemi Industriali). His main interests are related to testing and monitoring of MV rotating
machines technology, manufacturing, testing and monitoring.
He received [Link]. in Electrical Engineering from University of Trieste, Italy, in 2010.
He is member of CIGRE WG A1.42 "Survey on Hydro Generator Instrumentation &
Monitoring".
Alan McGuigan has had a long and diverse career in all aspects of the electrical supply
industry. He is past member of the Institute of Engineers, Australia.
His previous experience includes extensive experience as a Test and Commissioning
Engineer covering construction and maintenance of medium to large substations, thermal
and hydro power stations. This was followed by 10 years experience in roles of design,
system planning and construction in the distribution area.
Alan commenced with OMICRON Australia in January 2008. He is Manager of Australian and
New Zealand offices and also fulfills the role Application Engineer. In this role he provides
technical support, training and demonstrations for all Omicron primary test equipment. He
also organises special focus symposiums and workshops.
Colin Feely is an Asset Strategy Engineer with Powercor Australia Ltd, a privatised
Distribution Company operating in Victoria, Australia.
Colin has had over 30 years of engineering and management experience in the Electrical
Industry in Victoria and a 12 month consulting assignment in South Sumatra.
Colin holds an Electrical Engineering Diploma from the Gippsland Institute of Advanced
Education.
His preferred job description is to have fun playing with Plant.

INCREASED OPERATION RELIABILITY THROUGH

CONTINUOUS PD MONITORING IN THE JAVA-BALI REGION
Andrea Piccolo, Laurentiu-Viorel Badicu, Wojciech Koltunowicz
Omicron Energy Solutions
Lorenzweg 5, D-12099, Berlin, Germany,
[Link]@[Link], [Link]@[Link], [Link]@[Link]
Constant
Citra Wahana Sekar Buana,
constant@[Link]
Ugan Suganda, Wahab Winoto, Yohanes Mulyana, Wirawan
PJB (PT Pembangkitan Jawa-Bali)
[Link]@[Link], wahab@[Link], [Link]@[Link], wirawan1975@[Link]

ABSTRACT
PJB (PT Pembangkitan Jawa-Bali), the main power utility in the Java-Bali islands, Indonesia, embarked in
2011 on an ambitious project to equip all of its important generating plants with continuous on-line partial
discharge monitoring. At that time they were facing the fact that their equipment, installed mostly in the eighties
and nineties, was approaching the period of major maintenance needs and needed a reliable tool to prioritize
effort and improve reliability. Currently, PD monitoring systems are installed on 18 turbo-generators and on 8
hydro generators, and there are more to be installed in near future. This article describes the advanced hardware
and software features of these systems, including elimination of disturbances and separation/identification of
different types of insulation defects based on synchronous, multi-channel and multi-frequency techniques. Two
case studies are also included, the first being the case of Gresik PLTU (thermal power unit) and showing the
advantage of having a fully digital system, whose settings can be easily remotely adjusted. The second is that of
Cirata Unit 7, a typical showcase of how to proceed with the assessment of the stator winding conditions, when
clear PD activities are diagnosed by the monitoring system.
KEYWORDS: Generator, Condition Based Maintenance, Stator Insulation, Partial Discharge, Monitoring.
1.

INTRODUCTION

PJB is the main power utility in the Java-Bali islands. Thanks to its installed 6793 MW, distributed in six main
power stations (fig. 1), PJB is delivering yearly 32094 GWh to the Java-Bali grid, through 150 kV and 500 kV
HV lines.

Figure 1. Location of the power plants in Java-Bali islands

The base load generation is covered by the thermal power plants of Gresik, Paiton, Muara Karang and Muara
Tawar (fig. 2), while the 19% of the installed power is based on the hydro-power plants of Cirata and Brantas,
which cover the peak-load hours energy demand (7% of the yearly generated power).

Figure 2. PJB installed power per power plant location

Beside its power generation business, PJB develops and manages other diverse businesses related to power
stations in order to maximize the potential of the company. On top of that, PT PJB created business areas in
Operation and Maintenance, Engineering Procurement and Contracting, and entered a joint-venture to develop
new power stations as well as to manage plant service Operation and Maintenance business [1].
Considering the fact that most of its installed generating power was installed in the early nineties, with some
units even commissioned in the mid-eighties, PJB is continuously looking for enhancing its capabilities in
Enterprise Asset Management.
With reference to the popular survey of 1199 hydro-generators, carried out by the CIGRE Study Committee A1,
WG A1.10 in 2009 [2], it is well recognized that the principal cause of generator failure (57%) is because of
insulation damage. Considering an average life expectancy of 25-30 years for a generator, most of PJB rotating
electrical machines are approaching or are already in the critical period for major maintenance (refurbishment or
rewinding).
Aware of evidence that the peak of maintenance activity will probably be between the years of 2017 and 2020,
PJB decided in 2012 to extend its capabilities in insulation condition-based assessment by installing PD
monitoring systems on its generator fleet [3].
Since 2012, twenty one monitoring systems are operating, thirteen on turbo-generators (46% of the installed
thermal power) and eight on the generators at Cirata hydro-power plant(77% of the available hydro power).Thanks
to the positive experience with the above mentioned machine, other five system has been recently installed and
commissioned in Gresik Block 2 and Muara Karang.
This following paper presents the achievements and challenges after more than two years since the first PD
monitoring system was installed and commissioned.

2.
2.1

MAIN CONTENTS
Challenges in Partial Discharge Monitoring

Partial discharge (PD) monitoring is a widely accepted indicator of insulation condition for rotating machines.
PD measurements are specified and strongly recommended for off-line testing and on-line monitoring [4,5,6,7].
Nevertheless, extensive application of PD monitoring for Enterprise Asset Management is still a challenging
task based on the monitoring solution selected. Suggested requirements include:

x
x
x

The data acquisition and pre-processing stage must be flexible for installation to any kind of machine
(hydro generators, turbo-generators, motors) and environmental conditions (measurement noise,
distance of the sensors from the equipment under monitoring, et. al.);
The monitoring system must be capable of trending and storing significant data for long periods;
Remote access to the system must be granted to ease the access to the database;
The monitoring software package must be able to provide tools both for basic measurement and
advanced data analysis.

The introduction of fully digital systems that include the synchronous multi-channel PD measurement
technique allows sensitive measurements and accurate separation and classification of PD sources in a wide range
of rotating machines [8]. With reference to the PJB experience, the example of integrating the above-mentioned
technique within a server-based monitoring system/software is presented.
2.2

Monitoring System Lay-Out

PJB represents the case of a typical utility, with a mix of different generating units, which have been installed
and commissioned at different periods.
Each monitored unit is equipped with a set of three 1.1nF capacitive couplers (one per phase), installed inside
the Insulated Phase Bus and connected to the three phases (fig. 3). The capacity of the sensors, ten times higher
than the more popular 80pF, guarantee high sensitivity to PD signals even in a frequency range lower than 1 MHz.

Figure 3. Installation of capacitive couplers

The sensors are connected to a three-channel data acquisition and pre-processing unit, which is based on FPGA
technology (fig.4).

Figure 4. Installation of data acquisition and pre-processing unit

Each channel is equipped with a 20 MHz high-sensitivity analogue-to-digital stage (14bits resolution
and 130dB dynamic range the best in its class), which samples pulses at 65 MSsec. The digitized signals are
then sent to real-time FPGA based system that accomplish the following functions:
x
x
x

Digitally filters the pulses within a freely-selectable bandwidth, in order to reduce measurement noise;
Populates the Phase Resolved Partial Discharge (PRPD) patterns;
Generates the 3PARD map, in order to allow separation of noise and PD [3,8,9].

The actual advantage of having a fully-digital acquisition unit is that it can be easily customized with reference
to the type of machine and installation conditions (i.e. distance of the sensors from the equipment under monitoring,
measurement noise).
The dialogue with the data acquisition and pre-processing units (usually one per machine), which include the
timing periodic for acquisition and storage of the results, is managed by the server-based monitoring software.
The monitoring software includes the following main functionalities:
x
x
x
x

Application for real-time PD measurement;

Storage of results in a database;
Warning logic, based on the trend results;
Web-based graphical user interface for monitoring data management.

The advantage of a server-based system is to have a single remote access point for several machines that can be
easily reached via the worldwide web whenever requested (fig. 5).

Figure 5. Access to the server with monitoring software

In fig. 6, the lay-out of a typical four generators-installation is shown.

Figure 6. Lay-out of typical four-generator installation

2.3 CASE STUDIED

The selected cases studied show a practical application of insulation condition-based assessment within the
context of complex monitoring systems.
In order to highlight the flexibility of the monitoring platform for different types of rotating machines,
examples of turbo generators and hydro generators have been chosen respectively.
Turbo and hydro generators are explicitly different in their construction. Even between turbo generators there
are several technical solutions (i.e. cooling system and insulation technology), which can result in different PD
activity.
The direct comparison of results from different machines is not a meaningful way to evaluate data and should
eventually often leads to a wrong diagnosis. Instead, the possibility to compare data from similar machines, which
are stored in a common platform, is a powerful approach to achieve a reliable diagnosis and to build a solid inhouse know-how of the insulation conditions to efficiently support Enterprise Asset Management.
The first case that refers to Gresik PLTU (thermal power unit) shows the advantage of having a fully digital
system, whose settings can be easily adapted to environmental changes by remote adjustment of the monitoring
settings.
The second case that refers to Cirata HPP, represents instead a typical case of how to proceed with the assessment
of the stator winding conditions when clear PD activities are detected.
2.3.1 GRESIK - PLTU
Gresik power plant is the biggest generation facility of PJB and is located in the province of East Java,
approximately 20 km to the northwest of the city of Surabaya. The power plant covers a surface of 78 hectares
and is composed of a steam power plant (PLTU-600 MW of installed power), a gas power plant (PLTG-40 MW
installed power) and three combined cycle blocks (Block 1, Block 2, Block 3) with 1578,78 MW installed power
(fig. 7).
In total, all sixteen generators are to be monitored for PD. Thirteen of them already have the monitoring system
in full operation and on the remaining three generators, the PD monitoring system will be installed in 2015.
The PLTU historically represents the core of the Gresik PP facility and is composed of four steam turbines:
x
x

PLTU 1 and PLTU2, commissioned in 1981, are equipped with two twin 100 MW steam turbines
coupled to 13.2 kV - air cooled generators;
PLTU3 and PLTU4, commissioned in 1988, are equipped with two twin 200 MW steam turbines
coupled to 15.0 kV - hydrogen cooled generators.


Figure 7. Block 1 in Gresik PP

As mentioned before, the most efficient way to achieve a reliable insulation assessment of the stator insulation
is by comparing the results from similar machines. The graph below shows the combined trend from PLTU 1 and
PLTU 2

Figure 8. Combined PD data trend in PLTU 1 and 2

After more than one year since the commissioning, an increase of PD activity has been observed between
September and October 2013 in PLTU 1 (blue, yellow and orange trends). The trend has been zoomed in order to
investigate in detail the source of the steep increase in PD amplitude (fig. 9).

Figure 9. Combined PD data trend in PLTU 1 and 2 related to the time period Sept Oct. 2013

The increase is more evident in Phase S (yellow) and has a fluctuating behavior on all the three phases, increasing
in average since the 18th of October.
From the analysis of the recorded patterns, the cause of the increase has been identified in the appearance of a
source of external noise or disturbances that periodically appears in the PRPD patterns.
In order to restore the capability of the system to monitor PD exclusively, extra measurements have been
performed by a team of PD monitoring specialists, with the target of evaluating an adequate system fine tuning.
The results from the tests, carried out via remote connection between Germany (where the monitoring technical
support is based) and Indonesia, showed that the monitoring band-pass filter central frequency can be moved from
2 MHz to 7 MHz, allowing the system to reject the noise while preserving adequate sensitivity to PD pulses.

In order to preserve the possibility to compare of results from PLTU1 and PLTU2 and validate the settings
upgrade, the monitoring band-pass filter has been adjusted in both the machines. The results, showed in the fig.
10, clearly confirm, both for PLTU 1 and PLTU 2, that:
x
x
x
x

The capability to reject the external noise has been increased;

The sensitivity to the observed PD activity is preserved;
The levels of PD activity are similar in the two machines, within an amplitude range of 200-400pC;
The PD activities, investigated by means of patterns shape analysis, can be identified with regular
internal PD activity.
Original Settings
2 MHz central frequency

Resetting to
7 MHz central frequency

PLTU 1

PLTU 2

Figure 10. PRPD diagrams in PLTU 1 and PLTU 2 for two different central frequency of measurements: 2 and 7 MHz

2.4 CIRATA - Unit 7

PLTA Cirata is one of the largest hydroelectric power plants in Southeast Asia (1008 MVA) and the biggest
hydroelectric power plant in Indonesia (fig. 11).

Figure 11. Cirata Power Plant

PLTA Ciratas power house consists of two main blocks Cirata I, commissioned in 1988, and Cirata II,
commissioned in 1997. Each block is composed of 4x126 MW-16.5 kV hydro-generators with the same design
and provided by the same manufacturer. The main contribution of PLTA Cirata is to deliver energy during the
peak load hours and recover the Jawa-Bali electric system in case of black-out (Black Start-up PP).
Installation of monitoring systems on all the generators was completed at the end of 2012.
Since the first months of trending Cirata II-Unit 7, showed a steep increase of PD activity (fig.12).

Figure 12. PD trend in Cirata II-Unit 7

As all the four generators of Cirata II have the same design and were commissioned in the same period, they
can be considered comparable. The first action taken was to compare Unit 7 with its neighbour Unit 8.

Figure 13. CIRATA II PD trend in Unit 7 and in neighbouring Unit 8

From the trend diagrams of average charge for Unit 7 and Unit 8, it is possible to understand the deviation of
Phase S-Unit 7.
To confirm the activity in phase S, pattern acquisitions have been made periodically to identify the type of source
and to confirm the phase location. In fig. 14, the results for Unit 7 and Unit 8 are presented, where each pattern is
synchronized with its relevant phase voltage.

01/03/2013

12/04/2013

28/05/2014

Unit
8

Unit
7

Figure 14. PRPD patterns in Unit 7 and 8 acquired in different time

The next step of the evaluation aims to isolate the investigated PD source from the internal PD activity. This task
is accomplished by means of the 3PARD separation technology.
01/03/2013

12/04/2013

28/05/2014

PD
Source

Other
PD

3PARD
map

Figure 15. PRPD patterns in Unit 7 and 8 acquired in different time

The investigated PD source, identified by the selected cluster in the 3PARD, can be associated with surface
tracking located in phase S.
Considering the other sources separately, it is possible to have an indication of the machine from the point of
view of internal discharges, which can be considered uniform in the three phases and comparable with the values
recorded in Unit 8 (twin of Unit 7).
The stator inspection carried out on September 2013 confirmed the presence of surface PD activity in six slots,
belonging to the investigated phase S.

Slot 7

Slot 44

Slot 80

Slot 99

Slot 234

Slot 252

Figure 16. Evidence of surface PD activity in six different slots of Phase S

3. CONCLUSION

Continuous PD monitoring of stator windings is proving to be an essential diagnostic tool for conditionbased maintenance of large generators operated by PJB, in order to improve their reliability and useful
life.

The main key to performing appropriate diagnosis of the state of the insulation is to accurately separate
and identify different PD sources from external noise and disturbances, even when they are concurrent
and produce similar symptoms. To achieve this, state-of-the-art synchronous multi-channel evaluation
techniques are applied through an advanced automated system.

Worldwide accessibility and possibility to cross-evaluate the monitoring data of the generators of whole
power plants, stored in central databases, allows improved integration of generators PD monitoring
within the Enterprise Asset Management system.

The usefulness of the monitoring system has already been proven by the successful correct diagnosis of
a problem in the end winding area of a large hydro generator, and subsequently fully confirmed by a
visual inspection.

REFERENCES
[1] PJB Pengbakitan Java-Bali website [Link]
[2] CIGRE Technical Brochure 392 / Survey of Hydro Generator Failures- 2009 CIGR
[3] Constant Monitoring in Electrical Rotating Machines Case Studied CEPED 2013, Bali
[4] IEC 60034-27-2, edition 1.0, On-line partial discharge measurements on the stator winding insulation of
rotating electrical machines March 2012
[5] IEEE 1434-2000, IEEE Guide to the Measurement of Partial Discharges in Rotating Machinery
[6] R. Bruetsch, M. Tari, K. Froehlich, T. Weiers and R. Vogelsang - Insulation Failure Mechanisms of Power
Generators IEEE 2008
[7] C. Hudon, M. Belec and M. Levesque Study of Slot Partial Discharges in Air-Cooled Generators IEEE
Transactions of Dielectrics and Electrical Insulation Vol.15, No. 6; December 2008

[8] L.V. Badicu, W. Koltunowicz, M. Koch and A. Piccolo - Return of experience from continuous PD
monitoring of rotating Machines;ISH2013, Seoul
[9] W. Koltunowicz, R. Plath - Synchronous Multi-channel PD Measurements - IEEE Transactions on Dielectrics
and Electrical Insulation Vol. 15, No. 6; December 2008.


)'* +$('*
('
$"%" +)$ *
'!
%" +)$ %
'*,%+$('



(%

(

" "&")


 

7,!&0-,-31
3*2'!&,,#*

#130#+#,21



+(#'%#&
+),-*+.'#/
 
+*")$
)",&


$"'$#
5<53DB?>93C
=2

$25B5C
'954




<1EC
ECDB91




&#
17,!&0-,-31
+3*2'!&,,#*

+#130#+#,21
.0-4'"#

,#5
.-2#,2'*
',
*-!2'-,

'"#,2'$'!2'-,
,"
1#.02'-,
-$
+3*2'.*#

1-30!#1
,"
2&#'0
"'1!0'+',2'-,
$0-+

#62#0,*
,-'1#
',
40'#27
-$

11#21
&#
',"#.#,"#,2

- 1#04#01
*')#
!-3.*',%

!.!'2-01
31&',%
2.1
!.!'2'4#
-0
',"3!2'4#
1#,1-01
+"#
2&#
+3*2'!&,,#*


#4*32'-,
$#1' *#
-,

+-2-01
%#,#02-01
2&0##
.&1#
20,1$-0+#01
-0
1',%*#*',#

! *#
1712#+1
5'2&
!0-11 -,"',%
*',)1
&0##
"'$$#0#,2
+3*2'!&,,#*

#4*32'-,

2#!&,'/3#1
0#
.0-.-1#"

.&1#
+.*'23"#
0#*2'-,
"'%0+



.&1#
2'+#

0#*2'-,
"'%0+


,"

!&,,#*
$0#/3#,!7
0#*2'-,
"'%0+


&#

13!!#11$3*
..*'!2'-,
-$
**
2#!&,'/3#1
'1
.0#1#,2#"


 


P
02'*
"'1!&0%#1
!0-112*)
"'%,-12'!1
.22#0,
0#!-%,'2'-,

',2#0$#0#,!#
13..0#11'-,




 



@1BD91<
49C381B75
%
9C
1
3?=@<5H
@8IC931<
@85>?=5>1

259>7
1
3?>C5AE5>35
?6
<?31<
5<53DB931<
CDB5CC
3?>35>DB1D9?>
9>

D85
9>CE<1D9?>
)85
%
F1B95DI
1>4
9DC
E>CD12<5
CD?381CD93

2581F9?B
=1;5
%
=51CEB5=5>D
?>5
?6
D85
=?CD
496693E<D
D1C;

9>
8978
F?<D175
D5CD9>7
$>
D85
?D85B
81>4
%
=51CEB5=5>D
9C

1
G?B<4G945
1335@D54
=5D8?4
6?B
9>CE<1D9?>
4917>?C9C
1>4
1

B5AE9B54
@1BD
?6
1335@D1>35
D5CD9>7
?6
=?CD
+
1CC5DC


%C
1B5
5HDB5=5<I
61CD
3EBB5>D
@E<C5C
G8?C5
B9C5
D9=5
1>4

G94D8
45@5>4
?>
D85
49C381B75
DI@5
1C
1
3?>C5AE5>35
?6
<?31<

3?>49D9?>C
9>
D85
9>CE<1D9?>
45653D
DI@5
1>4
75?=5DBI

?<<?G9>7
D85
%
C97>1<
@B?@171D9?>
6B?=
D85
@<135
?6
9DC

?B979>
D?
D85
45D53D9?>
C9D5
/

0
D85
6?<<?G9>7
1C@53DC
81F5
D?

25
D1;5>
9>D?
133?E>D


)85
<?CC
?6
9>6?B=1D9?>
12?ED
D85
49C381B75
@B?35CC
9DC5<6

G9D8
9>3B51C9>7
49CD1>35
6B?=
%
?B979>


)85
9>3B51C9>7
9=@13D
?6
@E<C5
@B?@171D9?>
@1D8
?>
%

G1F56?B=


)85
CE@5B@?C9D9?>
25DG55>
49665B5>D
C97>1<C
57
=E<D9@<5

%
C?EB35
3B?CCD1<;
5HD5B>1<
9>D5B65B5>35C



)85
B54E354
9>6?B=1D9?>
?6
D85
=51CEB54
%
@E<C5
4E5
D?

D85
<9=9D54
21>4G94D8
?6
D85
3?E@<9>7
E>9D
1>4
D85
21>4
@1CC
69<D5B
B5CDB93D9>7
D85
9>6?B=1D9?>
D?
%
@E<C5
1@@1B5>D

381B75
F1<E5
1>4
D?
D9=5
?6
?33EBB5>35
G9D8
B565B5>35
D?

13
F?<D175
@81C5
?><I

)85
9>DB?4E3D9?>
?6
1
4979D1<
D538>9AE5
3B51D54
>5G

?@@?BDE>9D95C
D?
9=@B?F5
C5>C9D9F9DI
C97>96931>35
1>4

B5@B?4E3929<9DI
?6
%
=51CEB5=5>DC
2I
61B
5H35549>7
D85

31@129<9D95C
?6
?<45B
1>1<?7
%
=51CEB9>7
CICD5=C
(?
31<<54

%
69>75B@B9>DC
?B
@1DD5B>C
=1;5
EC5
?6
%
@81C5
1>4

")'&$(
&*
#"

$(!&
 
"
"
#&!

(#&
 

=17>9DE45
C8?G9>7
133E=E<1D54
CD?381CD93
%
@E<C5C
<9;5

R6B?J5>R
CD1D9CD93
(E38
@1DD5B>C
=1;5
1ED?=1D93
C?6DG1B5
%

61E<D
B53?7>9D9?>
651C92<5
?><I
6?B
C@539693
DI@5C
?6
%C
<9;5

3?B?>1
F?94
?B
CEB6135
49C381B75


"?45B>
4979D1<
%
CICD5=C
1@@<I
=E<D9381>>5<

CI>38B?>?EC
@B?35CC9>7
?6
%
C97>1<C
D81D
=1;5
CD1D9CD931<

1>1<IC9C
@81C5B5C?<F54
@1BD91<
49C381B75
@1DD5B>C
@E<C5

C5AE5>35
1>1<IC9C
=E38
=?B5
5669395>D
>?D
?><I
6?B

49C3B9=9>1D9?>
25DG55>
%
1>4
9>D5B65B5>35C
2ED
1<C?
6?B

61E<DDI@5
945>D96931D9?>
1>4
<?31D9?>
)85
13DE1<
4B1G213;C

1>4
?@@?BDE>9D95C
?6
=?45B>
4979D1<
%
=51CEB9>7
CICD5=
1>4

4979D1<
%
41D1
@B?35CC9>7
1B5
@B5C5>D54
9>
D85
@1@5B






!







!
0



!

)85
1>1<?7
%
=51CEB5=5>DC
CD1BD54
G9D8
D85
9>DB?4E3D9?>

?6
D85
'
45D53D?B
9>


D85
69BCD
AE1C92B?1421>4
%

9>CDBE=5>D
D
@1F54
D85
G1I
6?B
D85
CE335CC
?6
1@@1B5>D

381B75
=51CEB5=5>DC
9>
@
9>CD514
?6
+
1C
G9D8
'+

=51CEB5=5>DC
&E1C92B?1421>4
%
9>CDBE=5>DC
81F5
1
69H54

35>D5B
6B5AE5>3I
1>4
D859B
<?G5B
1>4
E@@5B
3ED?66
6B5AE5>3I

14:ECD54
9>
CD5@C
B5CE<D9>7
9>
1
21>4G94D8
6B?=


;J
E@
D?


;J
?=@1B54
D?
>1BB?G21>4
9>CDBE=5>DC
%
@E<C5

B5C?<ED9?>
G1C
C97>96931>D<I
9=@B?F54

)85C5
1>1<?7
%
9>CDBE=5>DC
ECE1<<I
3?>C9CD
?6
1
21>4@1CC

69<D5B
1>4
1
@51;
<5F5<
9>4931D?B
C
<?>7
1C
D85
21>4@1CC

69<D5B
5HDB13DC
%
@E<C5C
G85B5
C@53DB1<
45>C9DI
9C
3?>CD1>D

D85
@51;
?6
D85
69<D5B
B5C@?>C5
9C
@B?@?BD9?>1<
D?
D85
1@@1B5>D

381B75
?6
D85
%
3EBB5>D
@E<C5
)89C
6B5AE5>3I
4?=19>

   
0



Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


(%+,'(-$ /
'!

%+#
.' #)('(,*
,%+$ #''"%

"*,)"&"'+*

 

@B?35CC
9C
31<<54
AE1C99>D57B1D9?>
/0
)85
?ED@ED
?6
D85

21>4@1CC
69<D5B
G1C
3?BB5<1D54
D?
D85
13
F?<D175
@81C5
)85

69BCD
31D1<?7
6?B
%
C?EB35
B53?7>9D9?>
@E2<9C854
2I
'

9>


4B5G
851F9<I
?>
CE38
697EB5C
/ 0
%
CE@5B9=@?C54

?>
1
!9CC1:?EC
697EB5
9C
CD9<<
1
3?==?>
G1I
D?
49C@<1I
%

2581F9?B
G9D8
B5C@53D
D?
13
@81C5







!

)85
5H9CD9>7
1>1<?7
9>CDBE=5>DC
=145
CD?381CD93

@B?@5BD95C
?6
%
F9C92<5
2ED
CD1D9CD93C
3?E<4>SD
25

1@@B?13854
)?
6E<69<<
D85
45=1>4
?6
B9C9>7
C395>D9693

9>D5B5CD
=E<D9381>>5<
@E<C5
85978D
1>1<IJ5BC
G5B5

45F5<?@54
/0
%
@E<C5C
3?E<4
81F5
255>
C?BD54
1>4

3?E>D54
133?B49>7
D?
D859B
85978D
@E<C5
85978D
89CD?7B1=C

,85>
=9>93?=@ED5BC
1>4
%C
253?=5
1F19<12<5
6<5H92<5

C?6DG1B521C54
CICD5=C
B5@<1354
D85
=E<D9
381>>5<
@E<C5

85978D
1>1<IJ5BC
6?B
B5C51B38
>


D85
69BCD

3?=@ED5B9J54
%
=51CEB9>7
CICD5=
G1C
9>DB?4E354
D?
D85

@E2<93
/0

CEBF5I
?>
51B<I
45C97>C
9C
B5@?BD54
9>
/0
1>4

1>
9>45@D8
DB51D=5>D
?6
%
4979D1<
41D1
@B?35CC9>7
9C
79F5>

9>
/O
0
)8B55
@B9>39@<5
45C97>C
?6
4979D1<
%
=51CEB9>7

CICD5=C
31>
25
49CD9>7E9C854


%81C5B5C?<F54
DB149D9?>1<
45C97>
G9D8
1>1<?7
21>4
@1CC
69<D5B
3?=29>54
G9D8
@51;
45D53D?B
?B
@E<C5C81@9>7

1=@<9695B
C<?G

3?>F5BD5B
1>4
3?=@ED5B


)9=5B5C?<F54
G9D8
G94521>4
@B51=@<9695B
4979D1<
CD?B175

?C39<<?C3?@5
?B
61CD

3?>F5BD5B
1>4
3?=@ED5B


)9=5B5C?<F54
G9D8
G94521>4
@B51=@<9695B
61CD


3?>F5BD5B
B51<D9=5
@B5@B?35CC9>7
1>4
3?=@ED5B

)85
=19>
14F1>D175
?6
1
4979D1<
@81C5B5C?<F54
=51CEB9>7

CICD5=
9C
D85
31@129<9DI
D?
B53?B4
1@@1B5>D
381B75
1=@<9DE45

%
1>4
13
F?<D175
@81C5
1>7<5

?6
5138
C9>7<5
%

@E<C5
)89C
41D1
9C
EC54
D?
1@@<I
CD1D9CD93
5F1<E1D9?>

=5D8?4C
57
D?
75>5B1D5

?B
89CD?7B1=C
G9D8

B5C@53D
D?
%
1=@<9DE45
%
B5@5D9D9?>
B1D5
1>4
13
F?<D175

@81C5
97EB5



'%30#




1>4


%'%
@1DD5B>

)85
@E<C5C81@9>7
9=@13D
?6
D85
21>4@1CC
69<D5B
G1C

81B=<5CC
6?B
@81C5B5C?<F54
%
=51CEB9>7
9>CDBE=5>DC

G8938
G5B5
45C97>54
D?
=51CEB5
1@@1B5>D
381B75
=17>9DE45

*>6?BDE>1D5<I
D85
?B979>1<
%
@E<C5
3EBB5>D
G1F56?B=
G1C

3?=@<5D5<I
<?CD
1>4
G1C
B5@<1354
2I
D85
@E<C5
B5C@?>C5
?6

D85
21>4@1CC
69<D5B
?B
5F5>
G?BC5
2I
D85
?ED@ED
?6
D85
@51;

45D53D?B

)?
F95G
D85
?B979>1<
%
@E<C5
G1F56?B=
G94521>4
?B

E<DB1G94521>4
=51CEB9>7
CICD5=C
2531=5
>535CC1BI
/

0

>CD514
?6
21>4<9=9D9>7
69<D5BC
CE38
CICD5=C
EC5
G94521>4

@B51=@<9695BC
1>4
3?==5B391<<I
1F19<12<5
4979D9J5BC
G9D8

(C
C1=@<9>7
B1D5
1>4
1>
1@@B?@B91D5
1>1<?7
21>4G94D8



J
?B
=?B5
)85
C1=@<54
41D1
6B?=
D85
4979D9J5B
9C

DB1>C65BB54
D?
D85
3?=@ED5B
6?B
@?CD@B?35CC9>7
)85
=1:?B

2?DD<5>53;C
?6
D89C
CICD5=
1B5
D85
<9=9D54
41D1
DB1>C65B
C@554

1>4
D85
<9=9D54
B5DB9775B
B1D5
?6
D85
4979D9J5B

335<5B1D54
@B?7B5CC
9>
C?6DG1B5
1>4
81B4G1B5
?6
4979D1<

D538>9AE5C
=145
9D
@?CC92<5
D?
3?=29>5
3?>D9>E?EC

B53?B49>7
?6
%
41D1
G9D8
G94521>4
%
=51CEB5=5>DC

,94521>4
1>4
E<DB1G94521>4
%
=51CEB5=5>DC
B5AE9B5

8978
C1=@<9>7
B1D5C
6?B
4979D9J1D9?>
>
3?>C5AE5>35
D85

D9=56B1=5
6?B
3?>D9>E?EC
41D1
13AE9C9D9?>
9C
<9=9D54
2I
D85

C9J5
?6
D85
4979D9J5B
=5=?BI
)85B56?B5
4979D1<
G94521>4

%
CICD5=C
4979D9J5
@E<C5C
?><I
G85>
D85
9>@ED
C97>1<

5H3554C
D85
DB9775B
D8B5C8?<4
%B?35CC9>7
@5B6?B=1>35
?6

3?EBC5
9C
<9=9D54
2I
D85
B5DB9775B
B1D5
1>4
D85
CD?B175

31@129<9DI

531EC5
D9=5
9>
25DG55>
3?>C53ED9F5
%
@E<C5C
3?>D19>C

>?
B5<5F1>D
9>6?B=1D9?>
DB9775B9>7
?>
D85
@E<C5
9DC5<6
C55=C

D?
25
D85
1@@B?@B91D5
C?<ED9?>
D?
@B5F5>D
6B?=
4979D9J9>7
D89C

@5B9?4
2ED
DB9775B9>7
G9<<
G?B;
G5<<
?><I
6?B
8978
C97>1<D?
>?9C5
B1D9?
(#'
>
B51<
CICD5=C
8?G5F5B
9>D5B65B5>35

31>
31EC5
1
49<5==1
6
D85
D8B5C8?<4
9C
C5D
D?
8978
F1<E5C

D85
4979D9J9>7
CICD5=
G9<<
?><I
DB9775B
?>
<1B75
@E<C5C
>

3?>DB1CD
1D
<?G
D8B5C8?<4
9>D5B65B5>35
31>
DB9775B
D85

CICD5=
5@5>49>7
?>
D85
B5<1D9?>C89@
?6
%
1>4
@E<C5DI@5

>?9C5
B5@5D9D9?>
B1D5
<?G
D8B5C8?<4
31>
<514
D?
4979D9J5

=19><I
>?9C5
9>CD514
?6
%
@E<C5C

C
<?>7
1C
%
@E<C5C
1B5
=51CEB54
3<?C5
D?
D859B
?B979>

1>4
3?>C5AE5>D<I
G9D8?ED
C97>96931>D
49CD?BD9?>
E<DB1
G94521>4
G1F56?B=
%
1>1<IC5C
1B5
B5<1D54
D?
D85
@8IC93C

?6
D85
49C381B75
@B?35CC
/

O
0
9665B5>D
49C381B75C
57

3?B?>1
F?94
49C381B75
CEB6135
49C381B75
C8?G
49665B5>35C

9>
%
@E<C5
G1F56?B=
?><I
F9C92<5
G9D8
G94521>4

=51CEB9>7
CICD5=C
D
<?G
=51CEB9>7
6B5AE5>395C
8?G5F5B

@E<C5C
6B?=
49665B5>D
49C381B75
@B?35CC5C
253?=5

9>49CD9>7E9C812<5
1C
D859B
C@53DB1<
45>C9D95C
1B5
CD9<<

3?>CD1>D
"51CEB5=5>DC
?6
1@@1B5>D
381B75
F1<E5
133?B49>7

D?

 
1B5
21C54
?>
3?>CD1>D
C@53DB1<
45>C9DI
D?

@5B6?B=
3?BB53D
AE1C99>D57B1D9?>
?6
D85
%
@E<C5
3EBB5>D

>
31C5
?6
+
5AE9@=5>D
G9D8
3?=@<5H
DB1>C65B
6E>3D9?>

<9;5
57
DB1>C6?B=5BC
?B
B?D1D9>7
=1389>5C
D85
%

G1F56?B=
9C
C97>96931>D<I
16653D54
2I
@E<C5
@B?@171D9?>
1>4

3?>C5AE5>D<I
D85
G1F56?B=
<??C5
1<<
9>6?B=1D9?>
12?ED
D85

DI@5
?6
D85
%
C?EB35

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


)'* +$('*
('
$"%" +)$ *
'!
%" +)$ %
'*,%+$('



(%

(

" "&")





!








!

)85
14F1>354
4979D1<
%
=51CEB9>7
CICD5=
%CICD5=
G1C

45C97>54
D?
?F5B3?=5
D85
<9=9D1D9?>C
=5>D9?>54
9>
C53D9?>


)89C
CICD5=
9C
31@12<5
D?
@5B6?B=
3?>D9>E?EC
1>4

CI>38B?>?EC
=E<D9381>>5<
%
=51CEB5=5>DC
9>
B51<D9=5

/
0


!







)85
C9=@<96954
45C97>
?6
D85
%CICD5=
9C
@B5C5>D54
9>

97EB5


'%30#



4F1>354
B51<D9=5
4979D1<
%
=51CEB9>7
CICD5=

)85
CICD5=
9C
=?4E<1B
1>4
3?>C9CDC
?6
?>5
?B
=?B5

13AE9C9D9?>
E>9DC
1
6925B
?@D93
2EC3?>DB?<<5B
1>4
1
%

9>3<E49>7
=51CEB9>7
C?6DG1B5
)85
%
C97>1<C
1B5
69<D5B54

1=@<96954
1>4
4979D9J54
9>
B51<D9=5
@9@5<9>5

1B389D53DEB5
,9D8
1>
1=@<9DE45
AE1>D9J1D9?>
?6


29D

1>4
1
C1=@<9>7
B1D5
?6

"(C
1
CE2C1=@<5
D9=5

133EB13I
?6


>C
9C
13895F54
6?B
CI>38B?>?EC
=E<D9
381>>5<
=51CEB5=5>DC
)85
AE1C99>D57B1D9?>
9C

@5B6?B=54
2I
1
4979D1<
21>4@1CC
69<D5B
)85
35>D5B

6B5AE5>3I
1>4
D85
21>4G94D8
31>
25
6B55<I
38?C5>
9>
D85

B1>75
6B?=
43
E@
D?

"J
C5<53D12<5
69<D5B
21>4G94D8

6B?=

;J
E@
D?

"J
D?
1F?94
49CDEB21>35C
1>4
D?

B5138
1>
?@D9=1<
(#'
5F5>
E>45B
>?9CI
?>C9D5

3?>49D9?>C
?B
13
@81C5
CI>38B?>9J1D9?>
1C
G5<<
1C
6?B
13

G1F56?B=
49C@<1I
1>4
1=@<9DE45
B5149>7
1
C53?>4

3?>F5BD5B
4979D9J5C
D85
F?<D175
3EBF5
G9D8

29D

;(C







<<
41D1
13AE9C9D9?>
1>4
@B5@B?35CC9>7
1B5
@5B6?B=54
9>
D85

13AE9C9D9?>
E>9D
D?
7E1B1>D55
?@D9=1<
@5B6?B=1>35
9>
C@554

1>4
C97>1<
AE1<9DI

6?EB
@1B1=5D5B
F53D?B
6?B
5138
C9>7<5
%

@E<C5
9C
945>D96954
/"

%



(0
G85B5
"
9C
D85
13AE9C9D9?>

E>9D
>E=25B
%
D85
1@@1B5>D
381B75
F1<E5

D85
13
F?<D175

@81C5
1>7<5
1>4
(
D85
12C?<ED5
D9=5
>
@1B1<<5<
D85

9>CD1>D1>5?EC
F1<E5
?6
D85
13
F?<D175
9C
C1=@<54
5F5BI

C

<<
@1B1=5D5BC
31>
25
CD?B54
6?B
?66<9>5
@B?35CC9>7
G9D89>
D85

B5@<1I
=?45
?B
1<D5B>1D9F5<I
5H@?BD54
6?B
@?CD@B?35CC9>7

G9D8
"1D!12
>
?C39<<?C3?@5
=?45

"J
G9D8
C1=@<9>7

B1D5
?6

"(C
1>4
1
)21C54
C@53DBE=
F95G5B
D?
1>1<IJ5

9>@ED
C97>1<C
G9D8
6E<<
21>4G94D8
1>4
D?
<?31D5
%
2I
D9=5
4?=19>B56<53D?=5DBI
1B5
1<C?
9>3<E454
138
13AE9C9D9?>
E>9D

3?>D19>C
1>
9>D5B>1<
D5CD
75>5B1D?B
D?
5>12<5
3<?C54<??@
C5<6
D5CD9>7

 

C
1>
?@D9?>
1

@
31<92B1D9?>
@E<C5C
3?85B5>D
D?

D85
CICD5=
3<?3;
31>
25
75>5B1D54
D
31>
25
EC54
D?

45D5B=9>5
D85
6B5AE5>3I
B5C@?>C5
@B?@171D9?>
45<1I
1>4

1DD5>E1D9?>
?6
D85
9>F5CD971D54
CICD5=
G9D8
5H35<<5>D
(#'

4E5
D?
3?85B5>D
1F5B179>7
1C
G5<<
1C
D?
C5BF5
1C
1
B5=?D5
3?>DB?<<54
%
31<92B1D?B




!!

)85
CI>38B?>9C=
?6
%CICD5=
G1C
3853;54
9>
D85

<12?B1D?BI
1>4
D85
=51CEB54
E>35BD19>DI
?6
D9=5
6?B
=E<D9
381>>5<
CICD5=
?6

13AE9C9D9?>
E>9DC
G1C
9>
1
B1>75
?6

?><I


>C

>
C9D5
4EB9>7
16D5B9>CD1<<1D9?>
D5CD
?6
1

<1B75

;+
-!%
312<5
CICD5=
6?B
D85
69BCD
D9=5
1


;=
<?>7
-!%
312<5
CICD5=
G1C
CE2:53D54
D?
1>
13

F?<D175
D5CD
G9D8
C9=E<D1>5?EC
%
=51CEB5=5>DC
1D


3B?CC2?>49>7
:?9>DC
1>4

(
D5B=9>1D9?>C
?6
5138

C9>7<5<9>5
312<5
/
0
)85
31<92B1D9?>
@E<C5C
?6

@

G5B5
654
9>
D8B?E78
1>
5HD5B>1<
?ED4??B
D5CD
D5B=9>1D9?>

)85
31<92B1D?B
?@5B1D54
CI>38B?>?EC<I
G9D8
D85
%

=51CEB9>7
E>9DC
F91
?@D931<
6925B
2EC
1>4
CE@@<954

3?85B5>D
C97>1<C
6?B
%
=51CEB5=5>DC
I
3?85B5>D

1F5B179>7
D85

@
31<92B1D9?>
@E<C5C
1>4
D859B
@1BD<I

B56<53D9?>C
1D
D85
>5HD
:?9>D
31>
3<51B<I
25
49CD9>7E9C854

6B?=
D85
213;7B?E>4
>?9C5
<5F5<
5F5>
?F5B
1
49CD1>35
?6


;=
97EB5













56?B5
=E<D9381>>5<
%
=51CEB9>7
5AE9@=5>D
2531=5

1F19<12<5
D8B55@81C5
1@@1B1DEC
G5B5
D5CD54
2I
EC9>7
1

=E<D9@<5H5B
D?
3?>>53D
D85
3?E@<9>7
45F935C
1D
5138
@81C5

?>5
16D5B
1>?D85B
D?
D85
%
=51CEB9>7
9>CDBE=5>D

569>9D5<I
D85
EC5
?6
1
=E<D9@<5H5B
9>8929DC
C9=E<D1>5?EC

%
=51CEB5=5>DC
?=@1B9C?>
?6
%
=51CEB5=5>D

B5CE<DC
D1;5>
G9D8
1
=E<D9@<5H5B
9C
D85B56?B5
>535CC1B9<I

21C54
?>
49665B5>D
D9=5
C57=5>DC
5C945C
D85

49C14F1>D175
D?
C@<9D
D85
1F19<12<5
D5CD9>7
D9=5
9>
25DG55>

D85
D8B55
@81C5C
B5CE<D9>7
9>
<5CC
%
41D1
6?B
CD1D9CD931<

1>1<IC9C
=E<D9@<5H5B21C54
%
=51CEB5=5>DC
CE665B

5C@5391<<I
6B?=
3B?CCD1<;

(5F5B1<
3?E@<9>7
=5381>9C=C
71<F1>93
31@139D9F5

9>4E3D9F5
5<53DB?=17>5D93
695<4
31EC5
3B?CCD1<;
9>

25DG55>
@81C5C
9>
D8B55@81C5
@?G5B
1@@1B1DEC
<9;5

75>5B1D?BC
?B
DB1>C6?B=5BC
%
3B?CCD1<;
CE@5B9=@?C5C

@81C5B5C?<F54
%
=51CEB5=5>DC
C?
=1;5C
%
@1DD5B>

9>D5B@B5D1D9?>
C?=5D9=5C
F5BI
496693E<D
?>C5AE5>D<I
%

3B?CCD1<;
G1C
=19><I
B571B454
1C
C?EB35
?6
9>D5B65B5>35
O

G85>
=51CEB54
G9D8
1
=E<D9@<5H5B


>
3?>DB1CD
D85
B5CE<DC
?2D19>54
G9D8?ED
=E<D9@<5H5B

1@@<I9>7
D85
CI>38B?>?EC
=E<D9381>>5<
CICD5=
C8?G
3<51B

5F945>35
D81D
3B?CCD1<;
56653DC
3?>D19>
9=@?BD1>D
9>6?B=1D9?>

?>
%
@E<C5
?B
=?B5
3?==?>
?>
C97>1<
@B?@171D9?>

?@5>9>7
>5G
CDB1978D
6?BG1B4
=5D8?4C
D?
B5=?F5
49CDEB21>35C

1>4
=?B5?F5B
D?
3<51B<I
49CD9>7E9C8
25DG55>
=E<D9@<5
%

C?EB35C
E5
D?
49665B5>D
C9D5C
?6
?B979>
%

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.

 


(%+,'(-$ /
'!

%+#
.' #)('(,*
,%+$ #''"%

"*,)"&"'+*

'%30#




853;
?6
D85
%
=51CEB9>7
CICD5=
@5B6?B=1>35
O
@B?@171D9?>
?6
31<92B1D9?>
@E<C5
1<?>7

;=
?6

;+
-!%
312<5
<9>5

@E<C5C
1C
G5<<
1C
@E<C5DI@5
9>D5B65B5>35C
G9<<
@B?@171D5
?>

49665B5>D
@1D8C
D?
D85
45D53D?BC
?6
D85
=E<D9381>>5<
%

=51CEB9>7
CICD5=
5D53D54
@E<C5
G1F56?B=C
G9<<

3?>C5AE5>D<I
25
49665B5>D
C?
B5<1D9?>C
?6
%
=17>9DE45C
1>4

D9=5C
?6
1BB9F1<
G9<<
25
E>9AE5
6?B
5138
%
C?EB35
>
75>5B1<

D85C5
E>9AE5
B5<1D9?>C
5>12<5
49CD9>3D
C5@1B1D9?>
?6
49665B5>D

%
1>4
9>D5B65B5>35
@E<C5C
)?
5>CEB5
@B?@5B
3?BB5<1D9?>
?6

@E<C5C
D85
D9=5
6B1=5
?6
CI>38B?>?EC
%
45D53D9?>
C8?E<4
25

1C
C8?BD
1C
@?CC92<5
DI@931<<I
6B?=


D?

NC

)85
>5G
5F1<E1D9?>
D538>9AE5C
6?B
CI>38B?>?EC
=E<D9
381>>5<
%
=51CEB5=5>DC
G5B5
45C97>54
1>4
9>F5CD971D54

?66<9>5
6D5B
?@D9=9J1D9?>
D85I
G5B5
9=@<5=5>D54
9>D?
D85

%CICD5=
C?6DG1B5
6?B
?><9>5
1@@<931D9?>
)85
C385=1D93

4917B1=
?6
D85
5<12?B1D9?>
?6
D85
%
C97>1<
9>C945
D85
CICD5=

E>9D
6B?=
D85
C97>1<

3?>F5BC9?>
E@
D?
@E<C5
@1B1=5D5B

5HDB13D9?>
9>
381>>5<
CICD5=
9C
C8?G>
9>
97EB5


%E<C5
@1B1=5D5B
5HDB13D9?>
1<B514I
5H@<19>54
9>
C53D9?>


5>12<5C
B51<D9=5
41D1
@B?35CC9>7
B1D5C
E@
D?


H


@E<C5C

@5B
C53?>4
D
D89C
C@554
C5<53D54
@E<C5
@1B1=5D5BC
?6
D8B55

49665B5>D
13AE9C9D9?>
E>9DC
31>
25
3?BB5<1D54
1>4
49C@<1I54
9>

49665B5>D
5F1<E1D9?>
4917B1=C
%'
%)'
1>4
'

<<
D85C5
4917B1=C
C8?G
49665B5>D
@E<C5DI@5
C?EB35C
9>

C5@1B12<5
3<ECD5BC
1>4
5138
3<ECD5B
9C
C5<53D12<5
6?B
C8?G9>7

D85
%'%
@1DD5B>
?6
?>5
C9>7<5
%
C?EB35
?><I






 





0




)85
%'
G1C
?B979>1<<I
45C97>54
6?B
5F1<E1D9?>
?6

D8B55@81C5
%
=51CEB5=5>DC
/
0
%'
1@@<931D9?>

B5AE9B5C
D8B55
9>45@5>45>D
%
?2C5BF5BC
<9;5
57
31@139D9F5

D1@C
?6
D8B55@81C5
DB1>C6?B=5B
2EC89>7C
?B
3?E@<9>7

31@139D?BC
3?>>53D54
D?
D85
D5B=9>1<C
?6
75>5B1D?BC
>
31C5
?6

%
=51CEB5=5>DC
?>
<?>7
D8B55@81C5
C9>7<5<9>5
312<5

CICD5=C
3B?CC2?>49>7
<9>;C
5>12<5
49CDB92ED54
9>4E3D9F5

D8B55@81C5
%
45D53D9?>
6?B
?><9>5
=51CEB5=5>DC
/
0

'%30#



"E<D9381>>5<
%
41D1
@B?35CC9>7
G9D8

%
13AE9C9D9?>
E>9DC

%'
F9CE1<9J5C
D85
B5<1D9?>C
1=?>7
1=@<9DE45C
?6
1

C9>7<5
%
@E<C5
9>
?>5
@81C5
1>4
9DC
3B?CCD1<;
75>5B1D54

C97>1<C
9>
D85
?D85B
DG?
@81C5C
%'
3?>CDBE3D9?>
2I
D8B55

CD5@C
9C
C8?G>
9>
97EB5



)85
69BCD
CD5@
97EB5


9C
D?
31<3E<1D5
D85
<?71B9D8=
?6

D85
12C?<ED5
F1<E5
#

% 
  ?6
1<<
D8B55
@E<C5C
?6
1
%

DB9@<5
)85>
5138
@E<C5
9C
DB1>C6?B=54
9>D?
1
F53D?B
B5<1D54
D?

9DC
@81C5
?6
?B979>
97EB5



Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


)'* +$('*
('
$"%" +)$ *
'!
%" +)$ %
'*,%+$('



(%

(

" "&")


9>1<<I
1449D9?>
?6
D85
D8B55
F53D?BC
B5CE<DC
9>
1
C9>7<5
4?D

@?C9D9?>
B5@B5C5>D9>7
?>5
%
DB9@<5
4?D
9>
97EB5



'%30#



?>CDBE3D9?>
?6

%'

)85
%'
81C
35BD19>
@B?@5BD95C


)85
4?D
@?C9D9?>
45@5>4C
?><I
?>
D85
<9>51B
1>4

45D5B=9>9CD93
DB1>C65B
6E>3D9?>
25DG55>
@E<C5
C?EB35
1>4

45D53D9?>
C9D5


)85
4?D
@?C9D9?>
9C
9>45@5>45>D
6B?=
@E<C5
=17>9DE45
1>4

13
F?<D175
@81C5
O
EC56E<
1<C?
6?B
43
%
D5CDC


%E<C5C
?6
?>5
@81C5
G9D8
CI==5DB93
3B?CCD1<;
D?
2?D8
?D85B

@81C5C
G9<<
25
3?>35>DB1D54
?>
?>5
?6
D85
%'
1H9C

DI@931<
6?B
31<92B1D9?>
?B
5H1=@<5
96
?>5
@E<C5
%

45D53D54
1D
@81C5
!
9C
89785B
D81>
9DC
CI==5DB93
3B?CCD1<;

B53?B454
1D
@81C5C
!

1>4
!
97EB5

%

%
%


%
D85>
D85
4?D
9C
49C@<1I54
5H13D<I
?>
D85
1H9C
!
)85

4?D
49CD1>35
6B?=
%'
?B979>
9>3B51C5C
G9D8
9>3B51C9>7

B1D9?
D?
9DC
3B?CCD1<;
@E<C5C


6
D85
3?E@<9>7
D?
D85
@81C5C
!
1>4
!
9C
>?D
CI==5DB93

57
%

%

%
D85>
D85
@?C9D9?>
?6
D85
4?D
9C

C896D54
D?G1B4
D85
1H9C
?6
D85
C53?>4
89785CD
@E<C5


%E<C5
DB9@<5C
G9D8
5AE1<
=17>9DE45
3?==?>
=?45

9>D5B65B5>35
G9<<
25
3?>35>DB1D54
3<?C5
D?
D85
%'

?B979>
%

%

%


)85
DB1>C65B
6E>3D9?>
25DG55>
@E<C5
C?EB35
1>4
45D53D9?>

@<135
9C
75>5B1<<I
9>F1B91>D
8?G5F5B
CE@5B9=@?C54
>?9C5

1>4
F1B91>35
?6
D85
49C381B75
G1F56?B=
G9<<
3B51D5

3<ECD5BC
9>CD514
?6
C9>7<5
4?DC
/
0

I
B5@5D9D9?>
?6
D89C
@B?354EB5
6?B
1
<1B75
>E=25B
?6
%

@E<C5C
%
C?EB35C
G9D89>
D85
D5CD
?2:53D
1C
G5<<
1C
?ED5B
>?9C5

1@@51B
1C
3<51B<I
49CD9>7E9C812<5
3?>35>DB1D9?>
?6
4?DC

3<ECD5B
9>
1
%'











0


)85
1BB9F1<
D9=5
?6
1
%
DB9@<5
45@5>4C
?>
D85
@B?@171D9?>

@1D8
6B?=
%
?B979>
D?
D85
D5B=9>1<C
?6
D8B55@81C5
1@@1B1DEC

/

 0


6
D85
%
C?EB35
9C
61B
6B?=
D85
45D53D9?>
C9D5
@E<C5

1=@<9DE45C
?6
1
%
DB9@<5
D5>4
D?
5AE1<
25DG55>
D85
@81C5C

1>4
G9<<
C8?G
3?==?>
=?45
@B?@171D9?>
?>C5AE5>D<I
61B

49CD1>D
%
C?EB35C
G9<<
25
49C@<1I54
3<?C5
D?
D85
?B979>
?6
D85

%'
)89C
<9=9DC
D85
31@129<9DI
D?
49CD9>7E9C8
25DG55>
CE38

C?EB35C
EBD85B=?B5
49CD1>D
@B?@171D9?>
G9<<
41=@
8978

6B5AE5>3I
3?=@?>5>DC
?6
D85
%
C97>1<C
<?G
@1CC
56653D
1>4

C?
B5AE9B5C
1
<?G
=51CEB5=5>D
6B5AE5>3I
D?
=19>D19>

 

>535CC1BI
C5>C9D9F9DI
D
<?G
=51CEB5=5>D
6B5AE5>395C
%

@E<C5C
@B?@171D5
9>
' #+
!#
DB1>C=9CC9?>
<9>5

@B?@171D9?>
D85B56?B5
@B?@171D9?>
49CD1>35
1>4
1BB9F1<
D9=5

1B5
49B53D<I
3?BB5<1D54
G8938
31>
25
EC54
D?
49CD9>7E9C8
61B

49CD1>D
%
C?EB35C

33?B49>7<I
%)'
81C
255>
45F5<?@54
D?
3?BB5<1D5

1BB9F1<
D9=5C
?6
1
%
DB9@<5
3?=@<5=5>D9>7
1=@<9DE45

3?BB5<1D9?>
9>
D85
%'
,9D89>
1>
14:ECD12<5
D9=5
6B1=5

(

1<<
%
DB9@<5C
1B5
945>D96954
97EB5

C8?GC
D85
<?793

3?>CDBE3D9?>
?6
%)'
)85
4917B1=
9C
49F9454
9>D?
C9H
5AE1<

C53D9?>C
9>
G8938
5138
?6
C9H
@?CC92<5
?B45B
3?=29>1D9?>C
?6

DB9@<5
@E<C5C
9C
49C@<1I54
H13D
@?C9D9?>
?6
D85
4?D
9C

45D5B=9>54
2I
1
F53D?B
G8938
>?B=

1>4
F53D?B
1>7<5

1B5

31<3E<1D54
133?B49>7
D?
5AE1D9?>C

1>4


'%30#



+9CE1<9J1D9?>
C57=5>DC
?6
%)'
6?B
C9H
49665B5>D
1BB9F1<
?B45BC

?6
D85
%
DB9@<5
@E<C5C

)85
F53D?B
>?B=

9>3B51C5C
G9D8
9>3B51C9>7
45<1I
25DG55>

D85
69BCD
1>4
D85
<1CD
@E<C5
?6
1
%
DB9@<5

,
B5@B5C5>DC
D85

?66C5D
1>7<5
?6
D85
@81C5
G85B5
D85
69BCD
@E<C5
?6
D85
%
DB9@<5

9C
45D53D54



 



 





1>4
(
1B5

=51CEB54
D9=5
?6
D85
69BCD
C53?>4
1>4
D89B4
=51CEB54
@E<C5C

G9D89>
1
%
DB9@<5


97EB5

C8?GC
D85
F9CE1<9J1D9?>
?6
49665B5>D
D9=5

49665B5>35C
25DG55>
D85
%
@E<C5C
9>
31C5
D81D
69BCD
@E<C5
9C

45D53D54
9>
@81C5
!

D85
C53?>4
9>
!
1>4
D85
<1CD
?>5
9>

@81C5
!

?B
9>CD1>35
96
D85
D9=5
49665B5>35
25DG55>
69BCD

DG?
@E<C5C
9C
F5BI
C=1<<
D85>
D85
B5CE<D9>7
4?D
G?E<4
25

49C@<1I54
25DG55>
D85
1H9C
G85B5
D85C5
@E<C5C
1B5
45D53D54

97EB5


$>
D85
?D85B
81>4
96
D85
<1CD
DG?
@E<C5C
?33EB

1<=?CD
C9=E<D1>5?EC<I
D85
B5CE<D9>7
4?D
G?E<4
25
49C@<1I54

?>
D85
1H9C
G85B5
D85
69BCD
@E<C5
9C
45D53D54
97EB5


>

D85
31C5
D81D
1<<
D8B55
@E<C5C
1B5
45D53D54
1D
1<=?CD
D85
C1=5

D9=5
D85
F53D?B
<5>7D8
G?E<4
25
J5B?


(

(




G8938
49C@<1IC
D85
4?D
1D
D85
?B979>
?6
D85
4917B1=
97EB5




Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.

 


(%+,'(-$ /
'!

%+#
.' #)('(,*
,%+$ #''"%

"*,)"&"'+*

'%30#

)5CD
39B3E9D
G9D8
DG?
%
C?EB35C

1>4


'%30#



+9CE1<9J1D9?>
?6
49665B5>D
D9=5
49665B5>35C
25DG55>
D85
%
DB9@<5

@E<C5C
G9D89>
D85
!

C57=5>D





!






0


)85
381>>5<
6B5AE5>3I
B5<1D9?>
4917B1=
3?BB5<1D5C

CI>38B?>?EC
=E<D9C@53DB1<
%
=51CEB5=5>D
D?
C5@1B1D5
%

6B?=
9>D5B65B5>35C
>
3?>DB1CD
D?
%'
1>4
%)'

=5D8?4C
'
5F1<E1D9?>
4?5C
>?D
>535CC1B9<I
B5AE9B5

D8B55
9>45@5>45>D
13AE9C9D9?>
E>9DC
/

0

"E<D9C@53DB1<
%
=51CEB5=5>DC
1B5
21C54
?>
@E<C5

C@53DB1
3?BB5<1D9?>
2I
1@@<I9>7
D8B55
49665B5>D
21>4@1CC

69<D5BC
49665B5>D
35>D5B
6B5AE5>395C
1>4
21>4G94D8

C9=E<D1>5?EC<I
)85
D8B55
21>4@1CC
69<D5BC
31>
25
DE>54
D?

1B51C
G9D8
<?G
9>D5B65B5>35
<5F5<
>
75>5B1<
D85
21>4@1CC

69<D5B
>?


C8?E<4
25
DE>54
D?
1
<?G
35>D5B
6B5AE5>3I
D?

5>12<5
3?>6?B=9DI
D?

  
1>4@1CC
>?

1>4

1B5

DE>54
D?
89785B
35>D5B
6B5AE5>395C
G85B5
@E<C5
@B?@171D9?>

56653DC
31EC54
1<B514I
49CD9>7E9C812<5
49665B5>35C
9>
C@53DB1<

B5C@?>C5C
?6
%
C97>1<
I
@B?@5B
38?935
?6
D85
21>4
@1CC

35>D5B
6B5AE5>395C
9D
253?=5C
@?CC92<5
D?
@5B6?B=
%

=51CEB5=5>DC
133?B49>7
D?
D85
CD1>41B4C
1>4
9>
D85
C1=5

=?=5>D
D?
B5=?F5
@B13D931<<I
1<<
CE@5B9=@?C54
9>D5B65B5>35

'
3?BB5<1D5C
D85
?ED@ED
?6
D85
D8B55
21>4@1CC
69<D5BC

5H13D<I
<9;5
%'
4?5C
G9D8
D85
@E<C5
1=@<9DE45C
?6
1
%

DB9@<5

)?
45=?>CDB1D5
D85
56653D9F5>5CC
?6
'
=5D8?4
1

C9=@<5
+
39B3E9D
G1C
38?C5>
)89C
39B3E9D
3?>C9CD54
?6
1

3?E@<9>7
31@139D?B

1
3?E@<9>7
45F935

DG?
>554<5

D?@<1>5
1BB1>75=5>DC

1>4

1>4
1>
9>4E3D1>35


97EB5

)85
13
D5CD
F?<D175
G1C
B19C54
E>D9<
3?B?>1

49C381B75C
G5B5
75>5B1D54
1D

1>4

97EB5


'%30#


(E@5B9=@?C54
%
@1DD5B>C
1D

1>4


'%30#


'
?6
>?9C5
1>4
%C
6B?=

1>4


97EB5C

1>4


45=?>CDB1D5
D85
31@129<9DI
?6
D85
=E<D9
C@53DB1<
%
5F1<E1D9?>
)85
CE@5B9=@?C54
%
@1DD5B>C
1B5

3<51B<I
C5@1B1D54



 






$><9>5
%
=51CEB5=5>D
1D
35>D5B
6B5AE5>3I
?6


"J

D
9C
?2F9?EC
D81D
%
@1DD5B>C
?6
CE38
8978
C9=9<1B9DI
31>
1>4
6B5AE5>3I
21>4G94D8
?6


"J
G1C
@5B6?B=54
?>

;+

81B4<I
25
C5@1B1D54
I
9>DB?4E39>7
D85
9>4E3D1>35
!
D85
DEB2?75>5B1D?B
97EB5


C8?GC
D85
%
@1DD5B>C
%'%
9>

%
@E<C5
G1F56?B=
1D

G1C
381>754
)85
%
45D53D9?>
D8B55
@81C5C
!

!
1>4
!
97EB5


C8?GC
%'
1>4

CICD5=
3?>>53D54
D?
D85
3?E@<9>7
45F935

G1C
DE>54
D?


%)'
4917B1=C
G85B5
65G
3<51B<I
49CD9>7E9C812<5
3<ECD5BC



1>4

"J
35>D5B
6B5AE5>395C
B5C@53D9F5<I
)85
31>
25
C55>
5138
B5@B5C5>D9>7
1
49665B5>D
C?EB35
?6
%
@E<C5C

'
3?BB5<1D9?>
?6
D85
B5C@?>C5
1=@<9DE45C
?6
D85C5
D8B55
G9D89>
D85
9>CE<1D9?>
CICD5=
?6
D85
=1389>5
?B
3<ECD5BC




69<D5BC
9C
C8?G>
9>
97EB5


G85B5
DG?
3<ECD5BC
B5@B5C5>D
1>4

6B?=
%'
1>4
6B?=
%)'
B531<3E<1D9?>
G1C

%C
6B?=

1>4

1>4
D85
D89B4
?>5
B5@B5C5>DC
D85
@5B6?B=54
D?
B53?>CDBE3D
D85
3?>F5>D9?>1<
%
@1DD5B>
B5<1D54

D?
D85
F?<D175
?6
@81C5
!

97EB5



>?9C5

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


)'* +$('*
('
$"%" +)$ *
'!
%" +)$ %
'*,%+$('



(%

(

" "&")


 

'%30#


%
@1DD5B>
1D

?><I

'%30#
 

%
@1DD5B>
1D

?><I

'%30#


%'
E@@5B
?>5
1>4
%)'
<?G5B
?>5

'%30#




'%30#



%'%
@1DD5B>C
9>
D8B55
@81C5C
?6
1
DEB2?75>5B1D?B

13;
DB1>C6?B=1D9?>
?6
=1B;54
3<ECD5BC
133?B49>7
D?
97EB5




!


)85
16D5B9>CD1<<1D9?>
D5CD
?6
D85

;=

;+
-!%

312<5
CICD5=
C55
C53D9?>

71F5
D85
?@@?BDE>9DI
D?
=51CEB5

%
?>

312<5
1335CC?B95C
C9=E<D1>5?EC<I
)?
3853;
D85

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


(%+,'(-$ /
'!

%+#
.' #)('(,*
,%+$ #''"%

"*,)"&"'+*



@5B6?B=1>35
?6
D85
9>CD1<<54
%
=51CEB9>7
CICD5=
5HD5B>1<

%
G1C
75>5B1D54
1D
1
G9B5
1DD13854
D?
D85
+
5<53DB?45
?6

D85
D5CD
D5B=9>1<
)85
%'%
@1DD5B>C
45D53D54
1D

3?>C53ED9F5

:?9>DC



1>4

1B5
C8?G>
9>
97EB5



'%30#
 

:?9>DC

%'%
@1DD5B>C
45D53D54
1D
D8B55
3?>C53ED9F5

;+
312<5

)85
%'%
@1DD5B>C
C8?G
5HD5B>1<
%
1C
G5<<
1C
9>D5B65B5>35

6B?=
D85
B5C?>1>35
3?>F5BD5B
EC54
D?
75>5B1D5
13
D5CD
F?<D175


)85
9>D5B65B5>35
6B?=
D85
B5C?>1>35
3?>F5BD5B
1@@51B54
1C

6?EB
F5BD931<
<9>5C
31EC54
2I
CG9D389>7
?6
D85
C?<94CD1D5

3?>F5BD5B
G9D8?ED
1>I
:9DD5B
9>
13
F?<D175
@81C5
2ED
G9D8

F1BI9>7
1=@<9DE45
)85
49CD1>35
25DG55>
DG?
:?9>DC
G1C

1@@B?H9=1D5<I


;=
C?
@E<C5C
>55454

C
D?
@B?@171D5
6B?=

:?9>D
D?
:?9>D


%?G5B
312<5C
13DC
<9;5
1
<?CCI
DB1>C=9CC9?>
<9>5
31EC9>7
>?D
?><I

<5>7D845@5>49>7
45<1I
2ED
1<C?
41=@9>7
<?G@1CC
2581F9?B

49C@5BC9?>
1>4
@1BD<I
B56<53D9?>
1D
5138
:?9>D
97EB5

)89C

5H@<19>C
D85
3?=@<5H
%'
@B5C5>D54
9>
97EB5


C8?G9>7
>?D

?><I
41=@9>7
?6
5HD5B>1<
%
1>4
B5C?>1>35
3?>F5BD5B
9>D5B65B5>35

9>
DG?
C5@1B12<5
3<ECD5BC


1>4

2ED
1<C?
CE@5B9=@?C54
3<ECD5BC

31EC54
2I
=E<D9@<5
B56<53D9?>
#5F5BD85<5CC
C5@1B1D9?>
?6
%'%

@1DD5B>C
G1C
13895F54
97EB5



'%30#




13;
DB1>C6?B=1D9?>
?6
=1B;54
3<ECD5BC
133?B49>7
D?
97EB5




335<5B1D54
@B?7B5CC
9>
C?6DG1B5
1>4
81B4G1B5
?6
4979D1<

D538>9AE5C
=145
@?CC92<5
D?
3?=29>5
3?>D9>E?EC
B53?B49>7

?6
%
41D1
G9D8
G94521>4
%
=51CEB5=5>DC


(I>38B?>?EC
=E<D9381>>5<
%
=51CEB5=5>DC
9>
B51<D9=5

31>
25
@5B6?B=54
G9D8
14F1>354
4979D1<
%
=51CEB9>7

CICD5=
D85
CI>38B?>9C=
?6

%
13AE9C9D9?>
E>9DC
?6

=E<D9381>>5<
CICD5=
9C
9>
1
B1>75
?6



>C
?><I


(I>38B?>?EC
%
=51CEB5=5>DC
1B5
?6
7B51D
9=@?BD1>35
6?B

C9=E<D1>5?EC
45D53D9?>
1>4
<?31D9?>
?6
%
9>
+
312<5C

1335CC?B95C
4EB9>7
16D5B9>CD1<<1D9?>
495<53DB93
D5CDC



%'
'
1>4
%)'
5F1<E1D9?>
=5D8?4C
5>12<5

3<51B
49CD9>3D9?>
?6
=E<D9@<5
%
C?EB35C
1C
G5<<
1C

CE@@B5CC9?>
?6
5HD5B>1<
9>D5B65B5>35C
1<<
D85C5
=5D8?4C

C8?G
49665B5>D
@E<C5DI@5
C?EB35C
9>
C5@1B12<5
3<ECD5BC
1>4

5138
3<ECD5B
9C
C5<53D12<5
6?B
C8?G9>7
D85
%'%
@1DD5B>
?6

?>5
C9>7<5
%
C?EB35
?><I


(5@1B1D9?>
?6
%
C?EB35C
F1CD<I
9=@B?F5C
%
@1DD5B>

1>1<IC9C
2I
B5=?F9>7
1>I
CE@5B9=@?C54
@1DD5B>
1>4
1<<
%

5F1<E1D9?>
=5D8?4C
G8938
B5<I
?>
C9>7<5
C?EB35

3?BB5<1D9?>
<9;5
D9=5?66<978D
?B
@E<C5C5AE5>35
1>1<IC9C

>?G
253?=5
EC56E<
5F5>
E>45B
3B9D931<
3?>49D9?>C

 

'%30#


%'
45B9F54
6B?=
%'%
@1DD5B>C

)85
1ED8?BC
G?E<4
<9;5
D?
D81>;
"B

=1>E5<
"B


$2B1<93
"B

1<;?>
1>4
B

'5D8=595B
6?B
D859B

3?>DB92ED9?>
9>
@B5@1B1D9?>
?6
D89C
@1@5B

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.


)'* +$('*
('
$"%" +)$ *
'!
%" +)$ %
'*,%+$('



(%

(

" "&")




/
0
.
"9>7
(
9B<1C5;1B1>
1>4
(
8?9

Q(9=E<1D9?>
?6

%1BD91<

9C381B75
%B?@171D9?>
9>
%?G5B
#5DG?B;R



)B1>C

%?G5B

5<9F5BI
+?<
 
@@
 
 

/0
&
(81?J85>
1>4
(
9B<1C5;1B1>
Q)85
(DE4I
?6
%
%B?@171D9?>

%85>?=5>?>
9>
%?G5B
#5DG?B;R

)B1>C
%?G5B
5<9F5BI
+?<



@@

O


 

/0

K>97
1>4
.
#
'1?
%1BD91<
9C381B75C
9>

<53DB931<

%?G5B

@@1B1DEC
+



/ 0
'
,

 
Q'53?7>9D9?>
?6
49C381B75CR
<53DB1
#?


@@


O



/0
'
1BD>9;1C
1>4

!5F9
Q
C9=@<5
@E<C5
85978D
1>1<IJ5B
6?B
@1BD91<

49C381B75
B1D5
=51CEB5=5>DCR

)B1>C
>CDBE=
"51C
+?<


@@


O 



/0

ECD9>
1>4
'
1=5C
Q$><9>5
4979D1<
3?=@ED5B
CICD5=
6?B

=51CEB5=5>DC
?6
@1BD91<
49C381B75C
9>
9>CE<1D9?>
CDBE3DEB5R

<53DB

>CE<
"17
+?<


#?




/0

,1B4
Q979D1<
D538>9AE5C
6?B
@1BD91<
49C381B75
=51CEB5=5>DCR


)B1>C
%?G5B
5<9F5BI
+?<

@@
O 



/0

(89=

(?B1781>
1>4
,
(95G
Q979D1<
C97>1<
@B?35CC9>7
1@@<954
D?

D85
45D53D9?>
?6
@1BD91<
49C381B75
>
?F5BF95GR

<53DB
>CE<

"17
+?<


#?

@@
O

 

/0

(89=

(?B1781>
1>4
,
(95G
Q@@<931D9?>
?6
4979D1<
C97>1<

@B?35CC9>7
1@@<954
D?
D85
45D53D9?>
?6
@1BD91<
49C381B75
%1BD


$@D9=9J54

3?>F5BC9?>R

<53DB
>CE<
"17
+?<


#?


@@

O

 

/
0

(89=

(?B1781>
1>4
,
(95G
Q5D53D9?>
?6
%
ED9<9J9>7
4979D1<

C97>1<
@B?35CC9>7
=5D8?4C
%1BD

$@5><??@
>?9C5
B54E3D9?>R


<53DB
>CE<
"17
+?<


#?


@@
O




/

0
%
"?BC89EC
QCC5C=5>D
?6
495<53DB93
457B141D9?>
2I
E<DB1
G94521>4

%
45D53D9?>R

)B1>C
95<53D
<53D
>CE<1D9?>
+?<

@@
 O
 



/
0

$;E2?
#
1I1;1G1
1>4

"1DCEC89D1
Q)85
B5<1D9?>C89@
25DG55>

@1BD91<
49C381B75
3EBB5>D
@E<C5
G1F56?B=C
1>4
@8IC931<
=5381>9C=CR


<53DB
>CE<
"17
+?<


#?

@@
O 
 

/
0
(
9=
1>4
)
"9JED1>9
Q81>75
9>
%
@E<C5
C81@5
G9D8
1759>7R
9>

1@1>

@@

O 



/
0
"%

O
@B?4E3D
2B956
1>4
C@5396931D9?>
$"'$#
5<53DB?>93C

<1EC
ECDB91
 

/
0

(38BK45B
(
(EDD?>
1>4
'
%<1D8
Q?==9CC9?>9>7
?6
1
+5BI
!?>7


;+
-!%
12<5
(ICD5=R
9312<5

+5BC19<<5C
B1>35
%1@5B

#?
 

/
0

%<1D8
'
%<1D8

=1>E5<
1>4
,
1<;>5B
Q(I>38B?>5

4B59@81C975
)59<5>D<14E>7C=5CCE>7
1>
!59CDE>7CDB1>C6?B=1D?B5>
F?B
$BD

1>4
9=
!12?BR
)138D17E>7
917>?CD9;
5<5;DB9C385B
5DB952C=9DD5<

5B<9>
5B=1>I
%1@5B



 

/
0
,
,59CC5>25B7

1B94
'
%<1D8

'5D8=595B
1>4
,
1<;>5B
Q$>
C9D5
%
45D53D9?>
1D
3B?CC2?>49>7
<9>;C
?6
+
312<5CR
'

(5CC9?>C
%1B9C
B1>35
%1@5B



 

/
0

52BEI>5
1>4
$
!5C19>D
Q
=5D8?4
D?
<?31D5
1
%
C?EB35
9>
1

G9>49>7
2I
@B?35CC9>7
C97>1<C
=51CEB54
1D
9DC
D5B=9>1<CR

D8
(

1>71<?B5
>491
@1@5B
>?





/
0

$2B1<93
,
1<;>5B
1>4
'
%<1D8
Q(5@1B1D9?>
1>4
9>49F94E1<

5F1<E1D9?>
?6
%
C?EB35C
9>
CD1D?B
G9>49>7
?6
B?D1D9>7
=1389>5C
2I

CI>38B?>?EC
=E<D9381>>5<
=51CEB5=5>DR
D8
(
!:E2<:1>1

(<?F5>91
%1@5B
)


 

/ 0
(
(381@5B

$2B1<93
'
%<1D8
1>4
,
1<;>5B
Q(I>38B?>?EC
=E<D9

D5B=9>1<
?>C9D5
%=51CEB5=5>DC
?>
@?G5B
DB1>C6?B=5BC
G9D8
1>

5>81>354
D9=5
49665B5>D91<
5F1<E1D9?>
=5D8?4R

D8
(
!:E2<:1>1

(<?F5>91
%1@5B
)
 

/
0
'
%<1D8
Q"E<D9381>>5<
%
=51CEB5=5>DCR

D8
(
59:9>7
89>1

%1@5B
 
 

/0

'5D8=595B
"
BL75B

B15D75
'
%<1D8
,
?<DE>?G93J


$2B1<93
1>4
,
1<;>5B
QH@5B95>35C
9>
?>C9D5
@1BD91<
49C381B75

=51CEB5=5>DC
1>4
@B?C@53DC
6?B
%
=?>9D?B9>7R
9>
?>49D9?>

"?>9D?B9>7

917>?C9C
"
 
59:9>7
89>1
%1@5B
"



-(!'#!&
-*23,-5'!8
G1C
2?B>
9>
<9G935

%?<1>4
9>


5
B5359F54
D85
"(
%8
1>4

(3
457B55C
9>
5<53DB931<
5>79>55B9>7
6B?=
D85

,1BC1G
*>9F5BC9DI
?6
)538>?<?7I
9>

 


1>4
 
B5C@53D9F5<I
B?=


D?


85
G1C
G9D8
>CD9DED5
?6
%?G5B
9>
%?<1>4

1C
1
B5C51B38
C395>D9CD
9>
D85
978
+?<D175

5@1BD=5>D
B?=


D?
 
85
G1C
G9D8

(
D1<I
G85B5
85
G1C
=19><I
9>F?<F54
9>

+
D5CD9>7
1>4
4917>?CD93C
?6
+
5AE9@=5>D
>

 
85
:?9>54
$"'$#
G85B5
85
9C

9>F?<F54
9>
=?>9D?B9>7
?6
+
5AE9@=5>D
5
9C
=5=25B
?6
'
,


1>4
'
4F9C?BI
,?B;9>7
B?E@

 R>CE<1D9>7
1C5CR
1>4
9DC
,C

5
9C
1ED8?B
?6
C5F5B1<
9>D5B>1D9?>1<
B5@?BDC




-,*"
*2&
G1C
2?B>
9>
5B<9>
5B=1>I
9>


5
B5359F54
D85
"(
1>4
%8
457B55C
9>

5<53DB931<
5>79>55B9>7
6B?=
D85
5B<9>
*>9F5BC9DI

?6
)538>?<?7I
9>


1>4

 
B5C@53D9F5<I

B?=


D?


85
G1C
G9D8
D85
5B<9>

*>9F5BC9DI
?6
)538>?<?7I
9>
5B=1>I
1C
1

B5C51B38
C395>D9CD
9>
D85
978
+?<D175

5@1BD=5>D
B?=


D?


85
G1C
G9D8

(
D1<I
G85B5
85
G1C
=19><I
9>F?<F54
9>
D85

69BCD

;+
@B5AE1<96931D9?>
D5CDC
?6
-!%
9>CE<1D54
312<5
CICD5=C
B?=


D?
 
85

G1C
8514
?6
D85
8978
F?<D175
<12
?6
%
5B<9>
>
 
85
:?9>54

$"'$#
G85B5
85
9C
9>F?<F54
9>
=?>9D?B9>7
?6
+
5AE9@=5>D
5
9C

=5=25B
?6
 


5B=1>
=9BB?B
3?==9DD55
?6

) 
1>4
?6

'



G?B;9>7
7B?E@C
1>4
)C
5
9C
1ED8?B
?6
C5F5B1<

9>D5B>1D9?>1<
B5@?BDC

Authorized licensed use limited to: Ronald Plath. Downloaded on December 22, 2008 at 17:21 from IEEE Xplore. Restrictions apply.

A-26

2012 IEEE International Conference on Condition Monitoring and Diagnosis

23-27 September 2012, Bali, Indonesia

Monitoring of 420 kV XLPE Cable System in

Underground Tunnel
D. Gieselbrecht, W. Koltunowicz, A. Obralic, T. Ritz

P. Christensen, B. Schneider, [Link] (*)

nkt cables a/s, Brndby, Denmark
(*) nkt cables, Cologne, Germany

OMICRON Energy Solutions

Berlin, Germany
[Link]@[Link]

[Link]@[Link]
insulation condition is necessary. This information can be
derived by monitoring certain diagnostic parameters during the
operation of the equipment. Consequently, continuous
monitoring is an essential tool for proper maintenance
management to guarantee the high level of asset reliability [2].
The monitoring system not only provides a set of excellent
components but also the know-how in data evaluation and long
term partnership, supporting any utilities decision over the
complete asset life cycle.
The concept of applied continuous monitoring system is
presented in Fig. 1. The signals from different sensors
measuring partial discharges, distributed temperature, oil
pressure in terminations and sheath voltage limiters are
acquired by multi-channel data acquisition units.

AbstractThis paper deals with continuous monitoring of

components of a 400 kV XLPE cable system placed in an
underground tunnel in South London. An advance concept to
assess and monitor the condition of the insulation of cable system
is described. Partial discharges are continuously monitored at all
joints and terminations and at the same time the system performs
measurement of oil pressure in 400 kV terminations as well as the
condition of all sheath voltage limiters located in the joint bays.
Special attention is given to the procedure of on-site acceptance
of monitoring system and to the system maintenance strategy.
Keywords-component;
discharge, monitoring

I.

HV/EHV

XLPE

cables,

partial

INTRODUCTION

In the UK, major cities suffer from heavily congested roads

and as a result the use of a new underground cable tunnels to
provide an environment to contain HV and EHV cable systems
has become ever more popular in recent years. The 10 km in
length and having a 3 m inner diameter tunnel between
substations at Beddington and Rowdown is part of National
Grids investment programme to meet the growing demand for
electricity in London and to reinforce electricity transmission
in the South East. The tunnel houses a new 400 kV, 2500 mm2
XLPE cable circuit. The longest cable sections are
approximately 1176 m, being the record for this voltage in the
UK to date. Installing cables in confined spaces such as
underground tunnels changed the logistics and the approach on
how these cable installation methodologies can be executed
safely and effectively to guarantee high system reliability [1].
A continuous monitoring system is applied to the cable
system. Partial discharges (PD) are continuously monitored at
all joints and terminations and at the same time the system
performs measurement of oil pressure in terminations and
check the condition of all sheath voltage limiters (SVLs)
located in the joint bays.
II.

Figure 1. Concept of monitoring system

In case of PD signals, the acquisition unit performs advanced
pre-processing of the raw data. The disturbances are removed
and main characteristics of the PD signal are determined. The
output of the data pre-processing is transferred to a server that
enables long-term data storage. Advanced intelligent preprocessing reduces the amount of data to adequate levels for
transmission over a communication network.

MONITORING SYSTEM CONCEPT AND COMPONENTS

The separation of PD sources and effective suppression of

external noise is achieved by the application of synchronous
multi-channel (3PARD) evaluation techniques. The 3PARD
diagram visualizes the relation among amplitudes of a single
PD pulse in one phase and its crosstalk generated signals in the
other two phases [2, 3].

A. Concept of the System

The majority of dielectric failures in HV and EHV XLPE
cable system can be assigned to the defects in the electrical
insulation system of joints and terminations. These defects
develop over cable system lifetime. In order to detect such
changes at an early stage, detailed information on the actual

978-1-4673-1018-5/12/$31.00 2012 IEEE

917

B. PD Acquisition System and Inductive Power Supply

Inductive high frequency current transformers (HFCT)
sensors are mounted on cross-bonding (CB) links are used to
detect PD directly at the accessories (Fig. 2). The PD
acquisition system consists of one four-channel, high-precision
and modular acquisition unit for each accessory, connected to a
data concentrator. One data concentrator collects monitoring
data from two or three acquisition units via optical fibers and
routes them to a server (Fig. 3). Pre-processing functions are
already performed in data concentrator like e.g.: bands pass
integration, gating, denoising and multi-source separation.
The active components of monitoring system need
electrical power for several processes. Beside the computer and
communication devices, also pre-amplifiers or signal
converters (electrical to optical, e.g.) located close to the
sensors have to be supplied with power. Power supply
solutions for HV cable monitoring systems in underground
tunnel face different constraints like e.g.; distributed sensing
and EMC related to transient overvoltages short circuit currents
in bad grounding conditions. Cable joints are usually locations
without installed local AC power supply. In existing cable
tunnels, due to safety reasons, no auxiliary cables are laid close
to the power cables. In contrast, AC power supply of
monitoring devices for cable terminals located in substations is,
in principle, available.
The Inductive Power Supply provides the necessary
electronics to supervise and manage the dc current delivered at
its output depending on the various HV cable current load
situations (Fig. 4). The PCBs and electronics elements are
filtered and optimized to avoid any disturbances of the PD
measurements close by.

Figure 4.

It is accepted practice in the UK to install HV and EHV

cable systems in tunnels with open air SVLs connected in
delta formation directly across the joint insulation flanges. A
major advantage of such a bonding configuration is the
avoidance of grounding points within the tunnel normally
required at each joint bay. An additional benefit is that the
voltage drop across the screen separation is smaller. The
disadvantage is that the rating of the SVLs must be higher and
therefore the physical size bigger. The access to the SVLs is
limited, so there is a requirement to continuously monitor their
status, such as:
x Normal operation (below inception voltage);
x SVL is short circuited when the conductive flashover trace
is generated through the SVL varistors;
x SVL is in open loop totally damaged (active elements
destroyed-exploded).
A SVL is a non-linear resistor and together with the parallel
cable screen at a cross-bonding link creates characteristic loop
impedance. This impedance will change according to the status
of the SVL and the operation of SVL monitoring system is
based on these changes. In the system, the PD acquisition units
located at the joint bay periodically inject signal pulses
(using their internal test generators). These signals propagate
through the HFCT sensors to the SVL impedance loop and

C. Sheath Voltage Limiters Monitoring

Sheath Voltage Limiters minimize the transient voltage
across the screen separation of cross-bonding joints during
switching or lightning transients and reduce the risk of
damaging these.

Figure 2.

PD acquisition system

Inductive Power Supply

Figure 3.

918

Schematic diagram of monitoring system

eventually arrive at the acquisition unit itself. The Fourier

transforms of injected and returned signals are calculated,
averaged and compared. No special sensors are required for
the scope and major part of evaluation of SVL monitoring data
is performed within the Monitoring Server. Differences can be
easily analyzed and distinguished by spectral processing
algorithm implemented in the Monitoring Server Software.
System principle has been preliminary tested and proven in the
wooden mock-up tunnel built at the nkt cables offices in
Brndby (Fig. 5 and 6) and finally checked after cable
installation in London tunnel.

Figure 7.

Software system architecture

TABLE I. DEFAULT TIME SCHEDULING FOR DIFFERENT MEASURED VALUES

Mode
Value

Permanent

Periodic

Partial Discharge

every 2-3 sec

for 1 every 1 h

Oil pressure at terminations

every 2- 3 sec

once every 1 h

SVL status

--

once every 8 h

During the permanent mode, the data are acquired every 23 seconds, compared with threshold values and in real time
displayed in graphical user interface. In case those data are
within normal margins, they will be colored in green. If the
values exceed thresholds for "warning" or "alarm" they are
colored in yellow or red accordingly.
Periodic measurements are initiated in equidistant time
spans, e.g. every hour. Duration of the periodic measurement
is normally 1 min. During this time span all mentioned scalar
values are calculated and PRPD (phase resolved PD) and
3PARD diagrams are acquired. These data are saved for later
post-processing and trend visualization. Unscheduled periodic
measurements are triggered in case of one or more measured
quantities exceeding the threshold level.
PD activity is displayed as PRPD for each phase/sensor
and for each separated PD source, respectively. Trend
diagrams of statistical parameters such as PD magnitude,
frequency of occurrence of PD pulses etc. are available.
Suitable filter options enable the user to select the data display
according to his specific interest. The user can set limits which
cause warning or alarm messages when exceeded. The
measured values are continuously compared with signal
levels. The measured quantities are colored according to their
value related to pre-seated warning or alarm threshold levels.
This means, if the detected PD level on any channel on one
assed exceeds a configurable threshold, the corresponding
value will be drawn in red (Fig. 8).
The SVL status is verified within the Server Software by
FFT-based spectral analysis of the signal injected to the XB
link loop impedance from PD acquisition unit. A statistical
model of normal SVL behavior is used for reference. The
model is constructed based on the SVL data collected during
starting limited time period of cable system operation with
different load conditions. SVL status analysis is based on
statistical comparison parameters between the model and
current SVL measurements which clearly distinguish normal
operation of SVL from short circuited and open loop SVL.

Figure 5. HFCT sensors at XB-links are sending and receiving impulses to

check the SVL status

Figure 6. Frequency signal response for different status of SVL

D. Server and Software Architecture

The server receives data for analysis, display, and storage.
The acquisition units are configured and remote-controlled by
the Monitoring System Software. The Software supports
remote access over TCP/IP. This allows operators to quickly
react to detected problems and access the stored data from any
remote location. The software is a highly modular, scalable
distributed system. Its system architecture consists of the
windows-based core part and the web-based control part (Fig.
7). The core part of the Monitoring Software is realized as
windows services and runs continuously without any direct
user interactions. The core system implements: collection and
persistence of measurement, data post-processing and analysis,
security tasks for data access and system operations, external
interfaces for data exchange over Ethernet or field bus.
E. Trend Analysis
The monitoring system provides data from each of the
acquisition units and oil pressure sensors in two time intervals:
permanent and periodic one (Table I).

919

maintenance schedule activity of the cable system. This would

require access to the system installations, cable tunnel,
manhole, shafts and other related substation facilities including
eventual outage of the cable system which requires respective
scheduling efforts. Software updates are foreseen. They would
include modifications, bug fixes, and potentially software
enhancements with new features. Regular checks and
evaluation of data values and trending for the acquired partial
discharges stored in the database of the monitoring system on a
per cable system accessory basis are also possible, if requested
by the system owner. In case of repetitively reported trending
alerts or alarms or specific PD events detected by the
monitoring system, PD Expert consultancy and support can be
requested by the system owner.

Figure 8. Graphic user interface

TABLE II. MAINTENANCE SCHEME OF MONITORIN SYSTEM

III. ACCEPTANCE OF THE MONITORING SYSTEM ON SITE

The monitoring system was routine tested in the factory and
later installed on-site (Fig. 9).

System
element

Maintenance scheme
Activity to be
performed

Periodicity

visual check

yearly

owner

functionality check

every 3 years

owner &
system provider

updates

every 3 years

system provider

data evaluation

periodic reports

system provider

expert consultancy

in case of PD
event

system provider

Hardare
Sofware

Responsibility

CONCLUSIONS

The monitoring system provider shall support the

asset owner in all stages of the monitoring project, from
system design to maintenance of the monitoring system in
service;
Figure 9. Installation of the system on site

Separation of PD sources and suppress of external

noise is performed by multi-channel evaluation techniques;

The site check of performance of the monitoring system was

performed according to the following steps:
x Step 1: verification of the functional readiness of the
measuring system and of the monitoring server
x Step 2: verification of the synchronous behaviour of the PD
measuring system
x Step 3: determination of PD impulse attenuation, damping
and dispersion along the cable system.
The following parameters were determined: PD impulse
attenuation, damping and dispersion along the cable, velocity
of the calibration signal in the cable, best frequency ranges for
PD measurements at all PD units (with highest signal to noise
ratio), PD detection path division factor for every chosen
frequency range.
IV.

The Inductive Power Supply provides the necessary

power to the monitoring equipment to supervise the 400 kV
cable system. The dc current delivered is depending on the
HV cable load situations, but even at very low load the
power is sufficient to run the monitoring system;

To verify the status of SVL, the system utilizes the

CB link loop impedance;

A modular, distributed monitoring software system

allows reliable long term storage of monitoring data and
provides access via web interface;
REFERENCES
[1]

MAINTENANCE OF THE SYSTEM AND CUSTOMER

SUPPORT

The services and maintenance scheme of the monitoring

system is presented in Table II. The installed system elements
i.e. sensors, acquisition units, power supplies, batteries and
fibre optical data transmission network elements are
periodically inspected and checked. These visual inspections
and functional checks include adjustments, repair or minor
maintenance activities. Such visual inspections are planned
once every three years in coordination and align with the

[2]

[3]

920

P. Christensen, J. Nielsen, A. Hanekom, D. Cheale, Improved

installation methodologies for EHV Cables in Undrground Tunnels,
paper A1.6, in proceedings of: 8th International Conference on
Insulated Power Cables (Jicable) , Paris-Versailles-France, 19-23 June,
2011.
W. Koltunowicz, A. Obralic, A. Belkov, D. Gieselbrecht,"Continuous
PD Monitoring of HV XLPE Cable Lines", paper B1.9, in proceedings
of: 8th International Conference on Insulated Power Cables (Jicable) ,
Paris-Versailles-France, 19-23 June, 2011.
W. Koltunowicz and R. Plath, 2008, "Synchronous multi-channel PD
measurements", IEEE Transactions on Dielectrics and Electrical
Insulation, vol.15, no.6, pp. 1715-1723, 2008.