Condition assessment
of a medium voltage
cable network by
insulation diagnostics
and failure statistics
Göran Semart
Master of Science Thesis
Stockholm, Sweden 2008
Condition assessment of a medium voltage cable
network by insulation diagnostics and failure
statistics
Göran Semart
Stockholm 2008
Thesis work
Kungliga Tekniska Högskolan
School of Electrical Engineering
Division of Electromagnetic Engineering
ii
iii
Foreword and Acknowledgement
This report is the result of the thesis work “Condition assessment of a medium voltage cable
network by insulation diagnostics and failure statistics”. This thesis work is a part of my
Master of Science Education in Electrical Engineering at Kungliga Tekniska Högskolan
(Royal Institute of Technology) in Stockholm. The thesis work was accomplished between
September 2007 and January 2008 under supervision of Hans Edin.
I would like to express gratitude to my supervisor assistant professor Hans Edin, for guidance
and for showing great enthusiasm for this project.
I would like to thank Olle Hansson at Fortum for proposing this project.
I also thank Anita Nilsson and Henrik Svensson with personnel at Ekerö Energi, for the
opportunity to perform cable-measurements on the Ekerö grid.
I express thanks to Ph. D. student Valentinas Dubickas at the division of Electromagnetic
Engineering, for performing the field-measurements at Ekerö.
I would like to thank Madeleine Due and Tony Boyle for helping me correcting the grammar.
And at last I would like to thank my brother Ove Semart for valuable comments regarding
disposition of the report.
Göran Semart
Stockholm 2008
iv
v
Abstract
In today’s power consuming society it is important for the power companies to be a reliable
distributor, especially in a non-regulated market as we have here in Sweden. Due to the last
years large power cuts in Sweden, for example the storm Gudrun, the demand on the power
companies to replace their overhead lines with underground cables has increased. As a result
of a lot more underground cables it will probably in the future be more important to use cable
diagnostic measurements. This is in order to discover and replace bad cables before a major
failure occurs, and in this way minimize the failure frequency.
Medium voltage XLPE-cables suffer from water tree degradation. Water tree degradation
means that water is penetrating insulation, not uniformly but in a structure that resembles a
tree. The water tree degradation could be locally or evenly distributed along the cable
depending on the cable construction. This phenomenon leads at the end to a cable breakdown.
The breakdown often occurs when a transient overvoltage enters the cable, for example when
the cable is reconnected after it has been out of service.
The main aim with this thesis was to perform cable diagnostic measurements on some of the
Ekerö Energi’s medium voltage cables, which had been diagnosed in 1998. The
measurements were analysed in order to estimate the condition of the cables. An evaluation
was done in order to try to find out if the measurements made in 1998 were trustworthy.
Qualitative models of different water trees were developed in order to demonstrate the
behaviour of these water trees. The instrument used in the field-measurements is called IDA
200 HVU, which is an instrument that measures the dissipation factor at variable low
frequencies at diverse voltage levels. The technique is developed at KTH. The report also
contains an overview of different commercial instruments for XLPE-cable diagnostic and
cable fault location instruments. This overview gives information on commercial available
diagnostic instruments, and which technique they use. In the report the theory for different
techniques used in diverse instruments are revealed. Failure statistics from Fortum’s
Stockholm grid and Ekerö’s medium voltage grid were analysed in this report in order to try
to decide if the cable faults were an issue for Fortum.
The measured cables at Stenhamra village in the Ekerö grid shows that some cables are aged
but not critical. Most of the cables are in a fairly good condition, even though the age of some
cables is almost 40 years. Ekerö Energi had only had three cable breakdowns related to cable
insulation the last four years. At least one of the cables was placed in clay soil and had
damage on the jacket. Further, there are very good condition to place cables in the soil at
Ekerö, due to that the soil consists of sand and gravel that has a draining quality. Another
reason that Ekerö Energi has avoided breakdowns by water tree deteriorated cables could
have been that they have used a lot of sand when placing the cables in the ground.
It is difficult to say if the measurements made in 1998 were trustworthy or not. What could be
said on the other hand, is that even though three of them were judged as old and four of them
judged as critical for nine years ago they are all still operating.
The commercial cable diagnostic instrument for XLPE-cables available on the market right
now are all off-line instruments. All instruments except one measure the dissipation
factor tan δ , at either a fixed low frequency or at a low frequency sweep or at power-line
frequency. The other technique used, is to measure the isothermal relaxation current. The
techniques used in these instruments are considered to be non-destructive for a voltage level
vi
up to U 0 . All fault location instruments in this report are based on time domain reflectometry,
TDR. The pre-location are made with two different methods, Time Domain Reflectometry
TDR and Arc Reflection Method ARM. The Arc Reflection Method is used to pre-locate high
resistance fault and the Time Domain Reflectometry is used to pre-locate low resistance
faults. The instruments are suited for low and medium voltage distribution cables.
The statistic from the Stockholm grid shows that the major number of faults is connected to
underground cables. The largest fault category for high voltage, low voltage and service
cables are the category lack of maintenance or worn out cables. The second large category for
all three voltage levels is the category unknown. The largest damages on the high voltage
system caused by failures in these two categories are connected to the XLPE-cables.
However, reading the statistic it is not possible to say if the XLPE-cable failures are
connected to insulation problems or something else. The fault frequency in Stockholm’s
XLPE-cables was 3,34 in 2005 and 3,76 in 2006 faults per 100 km cable and year. The total
fault frequency in Stockholm was 3,55, which is in the same size as the total fault frequency
in Göteborg, which was 3,58. The fault frequency in Stockholm’s PILC-cables was 1,15 in
2005 and 0,92 in 2006 faults per 100 km cable and year, which is slightly lower than
Göteborg’s failure frequency on PILC-cables.
The statistics from Ekerö Energi’s grid is more extensive. There has been access to statistics
reported to Darwin, Ekerö Energi’s own statistics and the failure reports, which the statistics
are based on. Extra information has also been available in some cases. The two largest fault
categories reported to Darwin was fabrication or material faults and unknown. When the
failure reports were examined, it showed that the seven faults reported under fabrication or
material fault could be divided into five different categories. Of these seven faults only three
were connected to insulation faults. The total fault frequency in Ekerö’s XLPE-cable grid is
1,99 faults per 100 km cable and year, which can be compared with the failure frequency
reported to Darwin, which is 2,41 faults per 100 km cable and year. However, the fault
frequency connected to insulation fault on Ekerö’s XLPE-cables is 0,4 faults per 100 km
cable and year.
Both statistics from the Stockholm and Ekerö grid shows that there is a need of more
categories inserted in the report system. The aim with inserting more categories is to refine
the statistic in order to get a more exact evaluation of what causes the interruptions. This
evaluation can then be used to direct the measure right, improve the failure statistic and
reduce the interruptions.
vii
Sammanfattning
I dagens energikrävande samhälle är det viktigt för energibolagen att vara en pålitlig eldistributör, speciellt på en avreglerad marknad som vi har här i Sverige. På grund av de
allvarliga strömavbrott vi har haft i Sverige de senaste åren, har kravet på elbolagen ökat att
ersätta luftledningar med kabel. Som ett resultat av allt mer markkabel kommer det troligtvis i
framtiden bli än mer viktigt att använda kabeldiagnostik. Diagnostiken används för att
upptäcka vatteninträde i kablarna, och i tid kunna ersätta dåliga kablar. För att på detta vis
minska felfrekvensen i nätet.
Mellanspänningskablar, PEX-kablar, lider av vatteninträde, eng. water treeing. Med
vatteninträde menas att vatten har trängt in i isolationen och bildat vattenträdsliknande
formationer. Vatteninträdet kan vara lokalt eller jämnt fördelat över kabeln beroende på
kabelkonstruktionen. Det här fenomenet leder förr eller senare till kabelfel i form av
genomslag (jordfel). Orsaken är oftast en transient överspänning, exempelvis när kabeln blir
återinkopplad efter att ha varit urkopplad.
Huvudsyftet med examensarbetet var att utföra kabeldiagnostik på Ekerö Energis
mellanspänningsnät. Detta gjordes på några utvalda kablar som diagnostiserades 1998.
Mätningarna analyserades för att kunna tillståndsbedöma kablarna. Ett försök att jämföra
mätningarna från 2007 med mätningarna från 1998 gjordes också. Kvalitativa modeller av
olika vattenträd togs fram för att ge en beskrivning av hur de olika vattenträden beter sig.
Instrumentet som användes vid fältmätningarna var ett IDA 200 HVU system. Instrumentet
mäter förlustfaktorn tangens delta som funktion av en låg frekvens vid olika spänningsnivåer.
Tekniken är utvecklad på KTH. Rapporten innehåller också en översikt av olika kommersiella
instrument för diagnostik av PEX-kablar och instrument för lokalisering av kabelfel. Denna
översikt informerar om vad det finns för olika kommersiella diagnostikinstrument på
marknaden, samt vilken teknik de använder. I rapporten redogörs också för teorin bakom
teknikerna som använd i de olika instrumenten. Avbrottsstatistik för Fortums Stockholms nät
analyserades för att försöka avgöra hur dominerande kabelfelen är.
Kabelmätningarna gjorda på kabelnätet i Stenhamra tätort på Färingsö (Ekerö) visar att vissa
kablar är åldrade men inte kritiskt åldrade. De flesta kablarna är i riktigt gott skick, trots att en
del kablar är nästan 40 år gamla. På Ekerö Energi har man bara haft tre kabelfel orsakade av
isolationsfel de senaste fyra åren. Där kabeln i åtminstone ett fall var förlagd i leråker och
hade en skada på manteln. På Ekerö är det mycket bra markförhållanden för kabelförläggning,
eftersom marken mestadels består av sand och grus som har dränerande egenskaper. En annan
anledning till att Ekerö Energi har undvikit kabelfel, orsakade av vatteninträde, kan vara att
man har haft tillgång till rikligt med sand vid kabelförläggningarna. Det har troligen bidragit
till att manteln på kabeln har klarat sig från skador, och därmed har vatteninträdet fördröjts
avsevärt.
Det är svårt att säga huruvida mätningarna gjorda på Ekerö 1998 är tillförlitliga eller inte. Vad
som å andra sidan kan sägas är, att trots att tre av kablarna var bedömda som gamla och fyra
av kablarna bedömda som kritiska för nio år sedan så är samtliga fortfarande i drift.
De kommersiella diagnostikinstrumenten för PEX-kabel som finns tillgängliga på marknaden
just nu är alla off-line instrument. Alla instrument utom ett mäter förlustfaktorn tan δ ,
antingen som en funktion av en fast låg frekvens eller ett frekvenssvep av låga frekvenser
eller vid fast nätfrekvens. Den andra tekniken som används är att mäta polariserings- och
viii
depolariseringsströmmar genom isolationen. Metoderna som instrumenten använder anses
vara icke förstörande för spänningsnivåer upp till U 0 . Alla fellokaliseringsinstrumenten i den
här rapporten använder sig av tidsdomän reflektometri, TDR. Förlokaliseringen görs med två
olika metoder, TDR och Arc Reflection Method, ARM. ARM används för att detektera fel
med hög resistans och TDR används för att detektera fel med låg resistans. Instrumenten är
avsedda för låg- och mellanspänningskablar.
Statistiken från Stockholm nät visar att de flesta felen är kopplade till jordkabel. Den största
felkategorin för högspännings-, lågspännings och servicekablar är kategorin bristande
underhåll eller uttjänta kablar. Den näst största felkategorin för alla tre spänningsnivåerna är
kategorin okänd. De flesta fel på högspänningsnätet orsakade av fel från dessa två kategorier
är kopplade till PEX-kablar. Genom att läsa statistiken är det dock inte möjligt att utläsa om
PEX-kabelfelen är kopplade till renodlade isolationsfel eller till något annat. Felfrekvensen på
Stockholms PEX-kablar var per 100 km och år 3,34 år 2005 och 3,76 år 2006. Den totala
felfrekvensen på Stockholms mellanspänningsnät är i samma storleksordning som Göteborgs
mellanspänningsnät. Felfrekvensen på Stockholms papperskablar var per 100 km 1,15 år 2005
och 0,92 år 2006, vilket är något lägre än Göteborgs felfrekvens på papperskabel.
Statistiken från Ekerö är mer omfattande. Här har det funnits tillgång till statistiken
rapporterad till Darwin, Ekerös egen statistik samt fel rapporterna som ligger till grund för
statistiken. Det har även funnits tillgång till extra information om vissa fel. De två största
felkategorierna rapporterade till Darwin var fabrikations- eller materialfel och okända fel. När
felrapporterna granskades visade det sig att de sju felen som var rapporterade under
fabrikations- eller materialfel kunde delas in i fem olika kategorier. Av dessa sju fel var bara
tre kopplade till isolationsfel. Den totala felfrekvensen på Ekerös PEX kabelnät är 1,99 fel per
100 km kabel och år, vilken kan jämföras med felfrekvensen rapporterad till Darwin som är
2,41 fel per 100 km kabel och år. Felfrekvensen kopplad till isolationsfel på Ekerös Pex
kabelnät är dock bara 0,4 fel per 100 km kabel och år.
Både statistiken för Stockholms och Ekerös nät visar att det behövs fler kategorier i
rapportsystemet. Vinsten med att lägga till fler kategorier är att förfina statistiken för att bättre
kunna utvärdera vad som orsakar avbrotten. Utvärderingen kan sedan användas för att rikta
åtgärderna rätt för att på så sätt förbättra felstatistiken och minska avbrotten.
ix
Table of contents
Foreword and Acknowledgement ............................................................................................. iv
Abstract ..................................................................................................................................... vi
Sammanfattning ......................................................................................................................viii
Table of contents ........................................................................................................................ x
1
Introduction ........................................................................................................................ 1
1.1
Background ................................................................................................................ 1
1.2
The aim of the thesis work ......................................................................................... 1
1.3
Disposition ................................................................................................................. 2
2
Theory ................................................................................................................................ 3
2.1
Introduction to the chapter ......................................................................................... 3
2.2
Frequency-domain, dissipation factor measurement.................................................. 3
2.2.1
Dissipation factor of a cable............................................................................... 3
2.3
Time-domain, recovery voltage measurement ........................................................... 6
2.3.1
Dielectric response ............................................................................................. 6
2.3.2
Polarisation and Depolarisation Currents........................................................... 7
2.4
Time Domain Reflectometry...................................................................................... 8
3
Method ............................................................................................................................. 10
3.1
Introduction to the chapter ....................................................................................... 10
3.2
Literature study of insulation diagnostic instruments .............................................. 10
3.2.1
Pax Diagnostics Idax-206 VAX-230................................................................ 10
3.2.2
Baur VLF Cable Test and Diagnostic System PHG TD/PD............................ 11
3.2.3
sebaKMT Cable Diagnostic System CDS........................................................ 12
3.2.4
sebaKMT VLF Test System............................................................................. 16
3.2.5
Tettex Instruments............................................................................................ 17
3.2.6
HV Diagnostics, HVA 30, HVA 30,5 and HVA 60 ........................................ 17
3.2.7
Omricon CPC 100 with CP TD1...................................................................... 18
3.2.8
Summary of diagnostic instruments for XLPE- and PILC-cables ................... 18
3.3
Literature study of fault location instruments .......................................................... 19
3.3.1
sebaKMT Surgeflex 15, 25, 32 ........................................................................ 19
3.3.2
sebaKMT Teleflex T30-E ................................................................................ 19
3.3.3
Megger PFL40-1500/2000 ............................................................................... 19
3.3.4
Hipotronics 5150 First response cable fault locator......................................... 19
3.3.5
Summary of fault location instruments ............................................................ 20
3.4
The measuring method used on the Ekerö grid in 1998........................................... 20
3.5
Method used to analyse field measurements from 2007 .......................................... 21
3.5.1
Criteria for analysis of IDA 200 HVU measurements ..................................... 21
3.5.2
Analysis of different cable designs .................................................................. 22
3.5.3
Dielectric response in terminations and cable joints........................................ 23
3.5.4
Temperature dependence of the response ........................................................ 23
3.6
Qualitative dielectric models of water deteriorated cables ...................................... 24
3.6.1
Equivalent model for bush tree deteriorated cables ......................................... 26
3.6.2
Equivalent model for bow-tie tree.................................................................... 28
3.6.3
Equivalent model for a leakage tree................................................................. 29
3.7
Method used to analyse the statistics ....................................................................... 30
3.8
Validity and reliability ............................................................................................. 31
4
Results .............................................................................................................................. 32
4.1
Introduction to the chapter ....................................................................................... 32
4.2
SINTEF’s analysis of the measuring instrument used on the Ekerö grid in 1998 ... 32
x
4.2.1
The report of SINTEF Energiforskning AS (SEfAS) ...................................... 32
4.2.2
The test procedure ............................................................................................ 33
4.2.3
Test results........................................................................................................ 33
4.3
The analysis of the measurement in 2007 ................................................................ 34
4.3.1
Measurements day one ..................................................................................... 35
4.3.2
Measurements day two..................................................................................... 42
4.3.3
Measurements day three................................................................................... 52
4.3.4
Results of the measured XLPE-cables made in 2007 ...................................... 59
4.4
Statistics from the Stockholm grid 2005 and 2006 .................................................. 60
4.4.1
Fault frequencies for cables at the high voltage level ...................................... 64
4.4.2
Consequences of a failure ................................................................................ 66
4.5
Statistics from the Ekerö grid 2004 to 2007............................................................. 67
5
Discussion ........................................................................................................................ 72
6
Conclusion........................................................................................................................ 75
Appendices ............................................................................................................................... 76
Appendix A .......................................................................................................................... 76
List of symbols and abbreviations........................................................................................ 76
Appendix B .......................................................................................................................... 78
Matlab code for the Bush tree models.................................................................................. 78
Appendix C .......................................................................................................................... 79
Pictures from measurements made on the Ekerö grid in 2007............................................. 79
References ................................................................................................................................ 80
Bibliography............................................................................................................................. 83
xi
1 Introduction
1.1 Background
Today is the demand of electrical power crucial for a well functioning society, the
consumption in Sweden the last ten years has varied between 143-150 TWh [1]. The last
years problems with blown down overhead lines and power cuts caused by storms, has hasten
the replacement of overhead lines in favour of underground cables.
It has been showed that medium voltage XLPE-insulated cables with insulation shields made
of graphite paint and paint, known as the first generation XLPE-cables, are suffering of severe
water tree deterioration. Water tree deterioration is water that has penetrated the semiconductor shield, which could be locally or evenly distributed along the cable. While the
water tree growth, the voltage breakdown strength decreases and will at the end lead to a
cable breakdown. The most common factor that leads to a cable breakdown is when a cable
with low voltage breakdown strength is energized in a reconnection.
There are three different circumstances that have to be fulfilled for water tree growth. Firstly,
soil that conserve water around the cable, like clay soil. Secondly, the presence of a high
electric field which means that this phenomenon only occur on medium and high voltage
cables. And at last, some kind of damage on the outer jacket insulation, which allows water to
enter the cable. Furthermore, newer generations of XLPE-cables suffer from water tree
deterioration, but the density of the trees is reduced and the speed of the tree growth is much
slower than for the first generation XLPE-cables.
With more and more underground cables in the grid, the demands for cable diagnostics will
increase in the future, in order to minimize the fault frequency. Replacing heavily water tree
deteriorated cables before the occurrence of a severe breakdown could minimize the fault
frequency. Nevertheless, a low fault frequency is important for the power companies’ image
to be a reliable power supplier. From an economical point of view, it is essential for the power
companies to minimize long-term faults to avoid having to pay back the grid fee.
1.2 The aim of the thesis work
The main aim with this thesis is to perform cable diagnostics on some of Ekerö Energi’s
medium voltage cables that had been measured in 1998. The measurements will be analysed
in order to estimate the condition of the cables. An evaluation will also be done in order to try
to find out if the measurements made in 1998 were trustworthy. Models of different water
trees will be developed in order to demonstrate the behaviour of these water trees. The
instrument used in the field-measurements is called IDA 200 HVU, which is an instrument
that measures the dissipation factor at variable low frequencies at diverse voltage levels. The
technique is developed at KTH. The report will also contain an overview of different
commercial instruments for XLPE-cable diagnostic and cable fault location instruments. In
the report the theory for the methods used in these instruments will be revealed. Failure
statistics from Fortum’s Stockholm grid will be analysed in order to try to decide if the cable
faults is an issue for Fortum.
1
Summary of the aim with this thesis:
•
•
•
•
•
Perform cable diagnostics on Ekerö Energi’s medium voltage grid.
Analysis of the diagnosed cables at Ekerö.
Attempt to evaluate if the measurements made in 1998 were trustworthy.
Overview of different commercial XLPE-cable diagnostic instruments and fault
location instruments.
Analysis of failure statistics at the Stockholm and Ekerö grid.
1.3 Disposition
An introduction with background and aim is presented in the first chapter. In the second
chapter the theory used behind different diagnostic instruments is presented. The third chapter
describes the methods used to reach the aim with this thesis work. Chapter four presents the
achieved results. Chapter five contains a discussion and chapter six a conclusion of the thesis.
Appendix A contains abbreviations and list of symbols. Appendix B contains Matlab code for
the water tree models. Appendix C contains pictures from the Ekerö grid measurements in
2007.
2
2 Theory
2.1 Introduction to the chapter
Dielectric spectroscopy is used in insulation diagnostic instruments, as a method to discover
deteriorated insulation. Dielectric spectroscopy can be performed either in the frequencydomain or in the time-domain. Dielectric spectroscopy in the frequency-domain is
measurements at a fixed frequency low or at a variable low frequency. Time is the limiting
factor in how low in frequency the measurement will be done. These measurements assume
that the signal has a periodic shape. The most common method, used for insulation
diagnostics in commercial available instruments, is to measure the dissipation factor, tan δ
[2].
Measurements in time-domain are made by measuring the voltages and the currents as a
function of time. The two most common measuring methods used in time-domain
spectroscopy are RVM return or recovery voltage measurement and measurement of the
polarisation and/or depolarisation current PDC [2].
To be able to repair a cable after a cable failure has occurred, it is necessary to discover the
location where the cable fault has occurred. There are several different commercial systems
available on the market. The measuring method used in most instruments is time domain
reflectometry, TDR.
2.2 Frequency-domain, dissipation factor measurement
Measurement of the dissipation factor is made in the frequency domain, and can be performed
by these three different ways.
Dielectric spectroscopy
Dielectric spectroscopy is when the capacitance and the losses, dielectric losses described by
dissipation factor tan δ , are measured as a function of frequency. This method measures the
capacitance and the losses in the test sample at different low frequencies at variable high
voltage levels up to U 0 . The voltage levels could be increased up to about 1.5 U 0 , if the test
sample seems to be in good condition [2].
VLF dissipation factor
Very low frequency, VLF, measures the dielectric losses, described by the dissipation
factor tan δ . The measurement is performed at a fixed low frequency at about 0.1 Hz and at
variable high voltages [2].
Dissipation factor at power line frequency
This method measures the dielectric losses at power line frequency, described by the
dissipation factor tan δ , at variable high voltages [2].
2.2.1 Dissipation factor of a cable
The dissipation factor tan δ is used to describe the value of dielectric losses. One method to
derive the expression for the dissipation factor tan δ is made as follows:
3
Cable insulation can be modelled by a resistance R (ω ) in parallel with a capacitor C (ω )
shown in figure 2.1. Where R (ω ) represents the lossy part of the dielectric and C (ω )
describes the lumped-circuit capacitance of the cable. When the cable is applied with a
voltage V, the total current I in the cable is the vector sum of the leakage current Il and the
charging current Ic [3].
I = I l + I c = V ( G (ω ) + jωC (ω ) )
(2.1)
I
Il
V
Ic
R(ω)
C(ω)
Figure 2.1 Model of cable insulation.
⎛
1 ⎞
G (ω ) is the conductance and it is defined as the reciprocal of R (ω ) ⎜⎜ G (ω ) =
⎟.
R (ω ) ⎟⎠
⎝
The relation between the capacitive current I c and the leakage current I l is shown in figure
2.2.
I
Ic=jωC(ω)V
δ
θ
Il=G(ω)V
Figure 2.2 Phase relationship between capacitive
and leakage currents in a cable dielectric.
4
The dissipation factor gives from figure 2.2: tan δ =
G (ω ) V
Il
1
=
=
I c ωC ( ω ) V ω C ( ω ) R ( ω )
(2.2)
There is a relation between the electrical parameters eq. 2.1 in the cable model in Figure 2.1
and the dielectric parameters of the cable insulation. The dielectrics parameters of the cable
insulation are the real permittivity ε ′ , the imaginary permittivity ε ′′ and the conductivity σ AC .
The cable capacitance is defined as: C (ω ) = ε r′C0
(2.3)
Where C0 is the capacitance in vacuum and ε r′ is the relative value of the real permittivity
ε′
where ε 0 is the permittivity of vacuum.
defined as:
ε0
In analogy the imaginary permittivity is introduced to describe the lossy part of the dielectric
R, so that eq. 2.1 may be expressed by:
I = jωεr C0V
(2.4)
Where εr is the complex relative permittivity and defined as:
⎛
σ ⎞
ε~r = ε r′ − j ⎜⎜ ε r′′ + DC ⎟⎟
ωε
⎝
0
⎠
Where ε r′′ is the relative value of the imaginary permittivity
ε ′′
given by:
and σ DC is the DC conductivity.
ε0
⎛
σ
The total current can be expressed as: I = jωC 0Vε r′ + ωC 0V ⎜⎜ ε r′′ + DC
ωε 0
⎝
The dissipation factor becomes:
⎞
⎟⎟
⎠
⎛
σ ⎞
σ
C0V ω ⎜ ε r′′ + DC ⎟ ε r′′ + DC
ωε 0 ⎠
ωε 0
I
⎝
tan δ = l =
=
Ic
C0V ωε r′
ε r′
(2.5)
(2.6)
The imaginary permittivity ε r′′ is referred to the loss factor and cannot be directly measured.
Instead the dielectric conductivity σ AC is measured. The conductivity is defined as the ratio
of the current density J to the electric field E . The capacitance for a parallel-plate structure
in vacuum is given by:
C0 =
A
ε0
d
(2.7)
5
Where d is the thickness of the dielectric and A is the electrode area. If C0 from eq. 2.7 is
used in eq. 2.5 and divided by the electrode area A , following current density expression is
given:
J = jωε r′
ε 0V
⎛
σ
+ ⎜⎜ ωε r′′ + DC
ε0
d
⎝
⎞ ε 0V ⎛
⎛
σ
⎟⎟
= ⎜⎜ jωε r′ + ⎜⎜ ωε r′′ + DC
ε0
⎠ d
⎝
⎝
⎞ ⎞ ε 0V
⎟⎟ ⎟
= ( jωε r′ε 0 + (ωε r′′ε 0 + σ DC ))E
⎟
⎠⎠ d
(2.8)
Where the real part of the expression is the AC conductivity term: σ AC = ωε ′′ + σ DC
σ AC − σ DC σ DC
+
ωε 0
σ
ωε 0
= DC
Rewritten eq. 2.6 gives tan δ =
ε r′
ε r′ωε 0
⎛ σ AC
⎞ σ
⎜⎜
− 1 + 1⎟⎟ = AC
⎝ σ DC
⎠ ωε r′ε 0
(2.9)
(2.10)
2.3 Time-domain, recovery voltage measurement
One method to quantify dielectric response of materials is the measurement of recovery or
return voltages, see figure 3.3. The measurement is made as follows. A constant DC voltage
charges the cable for a period of time. Then the DC voltage is switched off and the cable is
short-circuited for a short period of time. After that the recovery voltage is measured under an
open-circuit condition. The recovery voltage is caused by a residual polarisation, which is
built up in the cable. The cable is then discharged by the leakage resistance [4].
2.3.1 Dielectric response
If a dielectric material is influenced by an electric field E ( t ) , the current density J ( t ) of the
material can be written [5], [6]:
J (t ) = σ DC E (t ) +
∂D(t )
∂t
(2.11)
Where σ DC is the DC conductivity and D ( t ) is the electric displacement which is given as:
D ( t ) = ε 0ε ∞ E ( t ) + ΔP ( t )
(2.12)
Where ε 0 and ε ∞ are the permittivity of vacuum and the dielectric material. ΔP ( t ) is the
slow dielectric polarisation and it is connected to the dielectric response function f Dielectric ( t )
by the relationship shown below:
t
ΔP ( t ) = ε 0 ∫ f Dielectric ( t − τ )E (τ ) dτ
(2.13)
0
6
2.3.2 Polarisation and Depolarisation Currents
Measurement of polarisation and depolarisation currents in dielectric materials gives
information about the polarisation process in the test sample. By combining eq. 2.11, eq. 2.12
and eq. 2.13 the current density can be written:
J (t ) = σ DC E (t ) + ε 0 ε ∞
dE (t )
d
+ ε 0 ∫ f Dielectric (t − τ )E (τ )dτ
dt
dt 0
t
(2.14)
An external voltage U ( t ) generates the field strength E ( t ) for a homogenous material and
the current through the test sample with the geometric capacitance C0 can be written:
⎡ σ DC
i (t ) = C 0 ⎢
⎣ ε0
dU (t )
+
U (t ) + ε ∞
dt
t
⎤
d
(
)
(
)
−
f
t
τ
U
τ
d
τ
⎥
Dielectric
dt ∫0
⎦
(2.15)
The test sample has to be totally discharged before starting the measuring process. When that
is achieved a step voltage is applied with the subsequent characteristics:
⎧0
⎪
U ( t ) = ⎨U 0
⎪0
⎩
t<0
0 < t < tp
t > tp
(2.16)
During the applied voltage step U 0 (see figure 2.3) a polarisation current i p is built up in two
parts. One part is related to the conductivity of the test object and the other is related to
different polarisation processes inside the test object. The charging or polarisation current
through the test sample can be written as:
⎡σ
⎤
i p (t ) = C 0U 0 ⎢ DC + ε ∞δ (t ) + f Dielectric (t )⎥
⎣ ε0
⎦
(2.17)
Where σ DC is the DC conductivity and f Dielectric ( t ) is the response function and C0 is the
geometric capacitance.
7
Figure 2.3 Applied step voltage and the polarisation and depolarisation currents [5].
When the applied step voltage is switched off, the sample is short-circuited, which gives rise
to a depolarisation current id shown in figure 2.3. The depolarisation current can be expressed
as:
id ( t ) = −C0U 0 ⎡⎣ f Dielectric ( t ) − f Dielectric ( t + t p ) ⎤⎦
(2.18)
2.4 Time Domain Reflectometry
The TDR measurement instrument consists of a pulse or a step generator and a high-speed
oscilloscope. The pulse generator brings in a short voltage pulse with a fast rise time into the
cable with characteristic impedance Z 0 . Cable joints and other local variations like water tree
deteriorated areas have different characteristic impedances than the cable. These differences
in characteristic impedance give rise to a voltage reflection Vr , which is measured with the
high-speed oscilloscope. The voltage reflection coefficient is given by eq. 2.19 [7]
.
Γ=
Vr Z L − Z 0
=
Vi Z L + Z 0
(2.19)
Where: Vr is the reflected voltage wave.
Vi is the injected voltage wave.
Z L is the characteristic impedance of the irregularity in the cable.
Z 0 is the cables characteristic impedance.
8
The distance to the reflection is given by eq. 2.20 below [7].
l=v
tr
2
(2.20)
Where: l is the distance to the reflection.
v is the speed of the wave in the cable.
t r is the travelling time for the reflected voltage wave.
9
3 Method
3.1 Introduction to the chapter
To be able to understand the complex connection of water tree deteriorated cables and
diagnostics methods, the following methods were used to answer the questions: A literature
study of different commercial instruments was made, based on the theory in chapter two.
Field measurements at the Ekerö grid were performed with the IDA 200 HVU system.
Measurements on these cables were also done in 1998, comparisons were made with the new
measurements from 2007. The author also talked with Ekerö Energy’s service personnel
about the conditions for underground cables on Ekerö. Two methods were used in order to
analyse the measurements made on the Ekerö grid. Firstly, existing knowledge from Peter
Werelius Ph. D. thesis [2] was used. Secondly, quantitative models for different water trees
were developed. The models were made in order to show the behaviour of different water
trees. To be able to locate a cable fault, a fault location instrument is needed. A small
literature study was made to understand how they work (showed in chapter two) and a
research of commercial available instruments was made. Failure statistics from Fortum’s
Stockholm grid for years 2005 and 2006 and Ekerö’s medium voltage grid was evaluated.
3.2 Literature study of insulation diagnostic instruments
3.2.1 Pax Diagnostics Idax-206 VAX-230
Idax-206 VAX-230 is provided by Pax diagnostics and is a development of the IDA 200
HVU instrument. General Electric provided IDA 200 HVU. The measurement- and analysis
methods used in Idax-206 are based on research carried out at Kungliga Tekniska Högskolan
KTH (Royal Institute of Technology) in Stockholm. The technique used in the instrument is
to measure the capacitance and dielectric losses tan δ in the sample, at variable voltage (0200 Vˆ ) and frequency 0.0001 Hz - 1kHz . The maximum capacitance measurement range is
between 10 pF and 100 μ F [8], [9].
The instrument is designed for diagnostic measurement of electrical insulation that consists of
liquid and/or solid material. It can be applied on most high voltage equipment e.g. power
transformers, measuring transformers, bushings- and paper insulating cables. To detect nonlinear materials, like water treeing in XLPE-cables, a high voltage unit VAX-230 (formerly
known as IDA HVU) is needed. VAX-230 is an optional high voltage unit accessory used
with the system to increase the output voltage to 30 kV (21 kVRMS). The frequency range
using VAX-230 is 0.0001-100 Hz and I = jωCU < 30 mA [8], [9], [10].
The principle of the Idax-206 instrument is shown in figure 3.1. A Digital Signal Processing
(DSP) unit generates a test signal with the wanted frequency. The signal is amplified with an
internal amplifier and then applied to the sample. The voltage over and the current through the
sample are measured with high accuracy. For measurement input, Idax-206 uses the DSP unit
that multiplies the measurement signals with reference sine voltages, and then integrates the
results over a number of cycles. This method rejects noise and interference and allows Idax206, with high accuracy, to work with low voltage levels [8], [9].
10
Figure 3.1 Principle of the Idax-206 instrument [11].
3.2.2 Baur VLF Cable Test and Diagnostic System PHG TD/PD
The PHG (Programmable High Voltage Generator) is a VLF (very low frequency) cable test
instrument supplied by Baur. The PHG is available in two models, the PHG 70 and the PHG
80. The difference is the output voltage and the output current. The instrument can be
upgraded with dissipation factor measurement (TD) and partial discharge (PD) level
measurement with source localisation. The instrument can be used to test cables,
transformers, switchgears, sheaths and generators [12]. The PHG cable test instrument has
three different programmable test voltages:
•
•
•
Sinus wave, 38 kVRMS (PHG 70), 57 kVRMS (PHG 80)
Square wave, 57 kV/0.1 Hz (PHG 70), 80 kV/0.1 Hz (PHG 80)
DC, ± 70 kV (PHG 70), ± 80 kV (PHG 80)
The testing is made with a fixed frequency, which can be programmable between, 0.01 to 1
Hz. The maximum testable load capacity at different voltages is shown in the subsequent load
diagram in figure 3.2. Breakdown during the test is automatically detected and the instrument
is switched off.
11
Figure 3.2 Maximum testable capacitance versus test voltage level [12].
The PHG can be upgraded with a diagnostic instrument performing dissipation factor
measurement (TD, tan δ ). The testing is done with a sinus wave voltage at different voltage
levels and with a low frequency, 0.1 Hz. This type of measurement provides differentiated
global information of the ageing of PE/XLPE cables. The differentiation is made between
new, slightly and heavily water-tree damaged cables. The criteria that indicates water treeing
is if the tan δ is larger than 1.2 ⋅10−3 at 2U0 or that the difference of tan δ at 2U0 and U0 is
larger than 6 ⋅10−4 [11]. All essential actual data, like applied voltage level, the phase under
diagnostics, capacitive load, values of measured current, voltage and dissipation factor are
summarised in the TD main menu [12].
The PHG could also be equipped with partial discharge (PD) measurement and localisation of
the partial discharge source. This application is used on PE/XLPE and PILC-cables (Paper
Insulated Lead Covered cable). The partial discharge measurement provides information on
insulation diagnostics on a PILC-cable. The PD measurement also provides information about
whether there are installation errors or electrical trees at plastic cables that have not yet caused
breakdown. It can estimate if the dissipation factor measurement was influenced by partial
discharges in for example a joint [12].
3.2.3 sebaKMT Cable Diagnostic System CDS
The CDS (Cable Diagnostic System) for three phase diagnostics is manufactured by
sebaKMT. The CDS instrument is made for diagnostics on PE/XLPE and PILC-cables. The
techniques used in this instrument are Isothermal Relaxation Current, IRC, and Return
Voltage Method, RVM-analysis. The measurement principle, see figure 3.3, is made as
follows:
12
Figure 3.3 The measurement principle for the IRC and the RVM method [13].
The cable is charged for a period of time with a constant DC voltage. Then the voltage is
disconnected and the cable is short-circuited for a short period of time. When the short-circuit
is disconnected, the return voltage is measured in an open circuit condition for the RVMmethod. And the discharge/depolarisation current is measured for the IRC-method, see figure
3.4. Both methods measure the charging/polarisation current, which make it easier to detect
weaknesses and wet joints in PILC-cables and XLPE-cables [13].
Figure 3.4 Dielectric responses in time domain [13].
The IRC-diagnostic is a diagnostic method for deterioration ageing and status of PE/XLPE
insulated medium voltage cables. The analysis is based on the dynamics of the relaxation
process within the insulation. When the insulation becomes old or in one way or another
becomes worn-out, it will lead to a decrease in the residual strength in the polymer. These
changes are shown in the relaxation process [14], [15].
13
i(t)*t and idep(t) for swx00092.mat
-8
4
x 10
i*t [As]
3
2
1
0
0
10
1
10
2
10
Time [s]
3
10
4
10
-8
10
-10
i [A]
10
-12
10
-14
10
0
1
10
10
2
10
Time [s]
3
10
4
10
Figure 3.5 shows the result of an IRC measurement [29].
The measured current in the time domain showed in figure 3.5 above can be approximated by
a summation function showed in eq. 3.1.
3
i (t ) = I 0 + ∑ ai e −t / τ i
(3.1)
i =1
The equation 3.1 contains of the coefficients ai and the time constants τ i . τ 1 and a1 describes
the relaxation mechanism caused by bulk relaxation. The ageing indicates by a decrease of a 2
during the time constant τ 2 and an increase of a3 during the time constant τ 3 . The
relationship is described by eq. 3.2 to eq. 3.4 [14], [15].
t
Q(t ) = ∫ i (t )dt
0
(3.2)
3
Q(t ) = I 0 ⋅ t − ∑ aiτ i e
1
−
t
τi
3
+ ∑ a iτ i
1
14
I 0 ⋅ t = 0 τ 3 > 3τ 2 τ 2 > 3τ 1
τ
− 2
⎛
⎛ 1⎞
⎜
Q(τ 2 ) ≈ a1τ 1 + a 2τ 2 ⎜1 − ⎟ + a3τ 3 1 − e τ 3
⎜
⎝ e⎠
⎝
τ
− 2
⎛
⎜
Q(τ 3 ) ≈ a1τ 1 + a 2τ 2 1 − e τ 3
⎜
⎝
⎞
⎟
⎟
⎠
(3.3)
⎞
⎟ + a τ ⎛⎜1 − 1 ⎞⎟
⎟ 3 3⎝ e ⎠
⎠
τ
⎛
− 3
⎜1 − e τ 2
⎜
⎝
⎞ a τ ⎛ 1⎞
⎟ + 3 3 ⎜1 − ⎟
⎟ a1τ 1 ⎝ e ⎠
Q(τ 3 )
⎠
≈
A=
τ
Q (τ 2 )
− 2 ⎞
a 2τ 2 ⎛ 1 ⎞ a 3τ 3 ⎛⎜
1+
1 − e τ3 ⎟
⎜1 − ⎟ +
⎟
a1τ 1 ⎝ e ⎠ a1τ 1 ⎜⎝
⎠
aτ
1+ 2 2
a1τ 1
(3.4)
Here eq. 3.2 represents the space charge trapped during the time of τ 2 and the space charge
trapped during the time of τ 3 of the approximation function. Eq. 3.3 describes the correlation
and component a 0 is negligible for the evaluation. Eq. 3.4 gives the mathematical expression
of the ageing factor, the A-factor [14], [15].
The A-factor is correlated to discrete energy levels. The analysis is then based on the relation
between these energy levels, which makes the measurement independent of temperature and
external influence. The result is then compared with measurements stored in a database. The
measured result is classified in four different ageing groups [14], [15]:
1.
2.
3.
4.
Perfect
Mid life
Old
Critical
The presentation is made with a staple diagram with a probability in percent that the cable is
defined in the right ageing group. The program also gives a prognosis of the residual strength
of the cable defined as n ⋅ U 0 , where U 0 is given as the phase voltage [14], [15].
The RVM-analysis provides characteristic information on the ageing status and moisture
content of paper insulated cables. The voltage measuring range is 0-5000 V. The RVM
analysis is based on threshold values. The non-linear behavior of the return voltage opposed
to the increase of the charging voltage and the form of the measurement curve are analysed.
To determine the non-linearity in the initial slope of the return voltage a Qa factor is used. The
Qa factor is measured at two charging voltages, see figure 3.6. The initial slope for a dry
cable is 2.0 and decreases towards 1.5 for a high moisture cable [13].
15
Figure 3.6 Graph of RVM-method for three different paper insulated cable segments and evaluating p-factors [13].
The evaluation of the Quotient Qa is:
•
•
•
2.00….1.87
1.86….1.65
< 1.65
Dry cable
Moisture content
Wet cable
Another factor for evaluating the moisture content in a PILC-insulated cable is to measure the
p-factor, which is directly related to the de-composition of the cellulose. The classification of
the p-factor is based on the maximum value of the voltage U m [V] divided by the initial slope
s [V/s] times the time to the maximum voltage tm [s]. The cable is wet if the p-factor is above
0.2 [13].
3.2.4 sebaKMT VLF Test System
sebaKmt also offers a very low frequency VLF, test systems between 28-60 kV at 0.1 Hz. It
also contains a DC-test. The instrument test voltage is a generated cosine square wave and the
instrument is mainly made for detection of water tree deteriorated XLPE-cables. The testable
capacity varies with the test voltage, which could be seen in figure 3.7 [16].
Figure 3.7 Maximum testable capacitance versus test voltage level [16].
16
To evaluate the relative quality of the cable the instrument measures the leakage current. The
data could be stored and used to generate a test history for different measurements. The
instrument is also equipped with a breakdown detection, which means that the system
immediately will be switched off and the cable will be automatically discharged in case of a
breakdown [16].
3.2.5 Tettex Instruments
Tettex instruments provide three diagnostic instruments for dielectric losses measurement.
The instrument could be applied on transformers, rotating machines, bushings, cables,
capacitors and circuit breakers. All three instruments measure the dissipation factor tan δ at
local power-line frequency [17], [18], [19].
Tettex Midas
Tettex Midas is the most robust instrument used especially in harsh electrical environments
like substations and different field locations. The test voltage is programmable up to 15 kV.
The test frequency could be chosen between 15 and 400 Hz. The maximum test object
capacitance is 56 nF at 15 kV and 88 nF at 12kV. The Midas instrument has something called
temperature correction. That function is used to recalculate the measured results to a reference
temperature ( 20D C ), which makes it easier to compare earlier measured data. It also has an
analysis function, which makes it possible to compare latest measurements with stored data.
Pass and fail limits could be set absolute or relative [17].
Tettex 2820
Tettex 2820 has a programmable test voltage up to 15 A / ωcn V and the test frequency ranges
from 15 to 1000 Hz. This instrument has the same temperature correction as the Midas
instrument. It also has a signal analysis which analyses curve forms, spectrum and also shows
slow drifts and trending of the selected measuring value. Pass and fail limits could be set
absolute or relative [18].
Tettex 2877
Tettex 2877 has programmable test voltage between 50 V and 1.2 MV. The test frequency is
between 45 to 65 Hz and the maximum test object capacitance is 10 μ F [19].
3.2.6 HV Diagnostics, HVA 30, HVA 30,5 and HVA 60
The HVA is a universal high voltage test instrument provided by HV Diagnostics. The
instrument can be applied on cables, capacitors, switchgear, transformers, rotating machines,
insulators and bushings. The cable test is made with very low frequency, VLF. The
instrument is also provided with dual polarity, DC and cable jacket or sheath testing outputs
modes. To assist the operator the instrument automatically calculates the optimum frequency
for larger loads. If a breakdown should occur during testing the actual voltage is recorded and
displayed before the breakdown mode is activated. The output voltage can be programmable
up to 62 kVˆ (HVA 60) and the output frequency between 0.02 and 0.1 Hz and maximum test
object capacitance is 10 μ F [20].
17
3.2.7 Omricon CPC 100 with CP TD1
CPC 100 is a multifunctional test system, which can be equipped with an accessory for
insulation diagnosis, CP TD1. The instrument is provided by Omicron. The basic version of
the instrument can be used for voltage and current measurements of transformers, resistance
testing and testing of protection relays. The insulation diagnosis accessory measures the
dissipation factor tan δ . The voltage can be varied between 0 − 12kV and the frequency
between 15 − 400Hz . Maximum test object capacitance is 3 μ F . The dissipation factor
frequency sweep can vary between 15 − 70Hz [21].
3.2.8 Summary of diagnostic instruments for XLPE- and PILC-cables
Table 3.1 shows a summary of the commercial available diagnostic instruments for XLPEcables and PILC-cables. The table also shows the different techniques which the instruments
are using. All commercial insulation diagnostic instruments available on the market at this
time are off-line systems. There are two different techniques used to diagnose XLPE-cables in
different instruments. The first technique measures the dissipation factor tan δ and can be
divided into three different methods. The three methods measure the dissipation factor as a
function of either a low fixed frequency or at a low frequency sweep or at power-line
frequency. The reason to use a very low frequency is to bring down the size of the instrument.
The second technique measures the Isothermal Relaxation Current IRC. One instrument can
also diagnose PILC-cables and the technique used is the Return/Recovery Voltage Method
RVM.
Table 3.1 Summary of commercial available diagnostic instruments for XLPE-cables and PILC-cables.
Manufacturer
Pax Diagnostics
sebaKMT
Instrument
Idax-206 with
VAX-230
VLF diagnostic
system PHG TD
CDS
Technique
tanδ at a low variable frequency
(0.0001-100 Hz)
tanδ at a fixed low frequency
(Adjustable between 0.01-1 Hz )
IRC Isothermal Relaxation
Current
RVM Return Voltage Methods
sebaKMT
VLF
Tettex
instruments
Tettex Midas
tanδ at a fixed low frequency (0.1
Hz)
tanδ at a fixed local power
frequency (Adjustable between
15-400 Hz)
tanδ at a fixed local power
frequency (Adjustable between
15-1000 Hz)
tanδ at a fixed local power
frequency (Adjustable between
45-65 Hz)
tanδ at a fixed low frequency
(0.02-0.1 Hz)
tanδ at a variable frequency (1570 Hz)
Baur
Tettex 2820
Tettex 2877
HV Diagnostics
Omicron
HVA 30, HVA
30,5, HVA 60
CPC 100 with CP
TD1
18
Usability
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage PILC-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Diagnostics of medium
voltage XLPE-cables
Table 3.1 shows that all instruments except one measure the dissipation factor tanδ as a
function of a variable or a fixed frequency to diagnose XLPE-cables. The other technique
used in sebaKMT’s CDS to diagnose XLPE-cables is IRC Isothermal Relaxation Current
measurement. This instrument can also be used to diagnose PILC-cables, the method used
then is RVM Return Voltage Method. All instruments are considered to use non-destructive
measuring methods if the voltage is lower or equal to U 0 . However, a bad cable, although
good enough to be re-connected, can be damaged if the voltage is risen over U 0 .
3.3 Literature study of fault location instruments
To find out what kind of different commercial fault location instruments that are available and
which method they use, a literature study was done. The literature used in this thesis work is
literature, Ph. D. thesis, reports and leaflets.
3.3.1 sebaKMT Surgeflex 15, 25, 32
SebaKMT provides the Surgeflex systems. Surgeflex 15 and Surgeflex 25 are battery
powered fault location systems with 15 kV DC respectively 25 kV DC output voltages. After
identifying the type of fault the fault pre-location can start. Their are two different methods
used in this models, Time Domain Reflectometry TDR and Arc Reflection Method ARM.
The Arc reflection Method is used to pre-locate high resistance fault and the Time Domain
Reflectometry is used to pre-locate low resistance faults. The instruments are suited for low
and medium voltage distribution cables. The Surgeflex 32 has an output voltage of 32 kV DC
and it has except for the TDR and ARM also Impulse Current Method ICE and Voltage
Decay Method for prelocating the fault position. Another feature available on the Surgeflex
32 is Burning, which is a high voltage dc output with a high current used for easier and
quicker prelocation of unstable, flashing or high resistance faults [22], [23].
3.3.2 sebaKMT Teleflex T30-E
SebaKMT provides the Teleflex T30-E. The instrument uses the Time Domain
Reflectometry, TDR. The instrument is designed for low resistance faults in low and medium
voltage cables [24].
3.3.3 Megger PFL40-1500/2000
The PFL40-1500/2000 is provided by Megger and is a portable cable fault location and high
voltage test instrument. Megger PFL40-1500/2000 is an instrument with DC testing output up
to 40 kV. After identifying the type of the fault, pre-location of the fault position can be
decided. TDR is used to pre-locate low resistance faults, ARM and Proof/Burn is used for
unstable faults or high resistance faults. Another method is ICE and Voltage Decay Method
[25].
3.3.4 Hipotronics 5150 First response cable fault locator
The 5150 first response cable fault locator is provided by Hipotronics. The instrument uses
the digital ARC reflection Time Domain Reflectometry, TDR. The instrument is mobile with
a battery supply. The measured data can be downloaded from the instrument [26].
19
3.3.5 Summary of fault location instruments
Table 3.2 gives a summary of the fault location instruments revealed in this report. All fault
location instruments in this report are based on time domain reflectometry, TDR. The prelocation are made with two different methods, Time Domain Reflectometry TDR and Arc
Reflection Method ARM. The Arc Reflection Method is used to pre-locate high resistance
fault and the Time Domain Reflectometry is used to pre-locate low resistance faults. The
instruments are suited for low and medium voltage distribution cables.
Table 3.2 Summary of commercial available fault location instruments for low and medium voltage distribution cables.
Manufacturer Instrument
Technique Usability
sebaKMT
Surgeflex 15, 25, 32
TDR
sebaKMT
Teleflex T30-E
TDR
Megger
PFL40-1500/2000
TDR
Hipotronics
5150 First response
cable fault locator
TDR
Fault location, Burning (Surgeflex
32)
Fault location for low resistance
faults
Fault location and high voltage test
instrument with Proof/Burn
Fault location instrument
3.4 The measuring method used on the Ekerö grid in 1998
The instrument used at the last measurements in 1998 was Seba Dynatronics KDA-1. The
diagnostic method used in this instrument, to evaluate the integral ageing and deterioration
status of PE/XLPE insulated medium voltage cables, is the isothermal relaxation current
measurement. The instrument Seba Dynatronics KDA-1 is not commercial available anymore.
Prerequisite for cable measurements [4]
• Before starting the measurements the cable must be disconnected from the grid for
about four hours, depending on how heavy the load has been.
• The cable must have been earthed at least one hour before the start of the
measurements.
• The cable must not have been disconnected for a longer period of time, in this case the
cable needs to be re-connected to the grid for at least one week before starting the
measurements.
Before the measurements starts the phases and the shield of the cable has to be disconnected
to achieve the best result. After setup, see figure 3.8 of the instrument, the measurement starts
with the charging of the cable for about 1800s. The next step after the formation is to
discharge the cable over a load for about 5s. The last step of the measurement is to measure
the depolarisation current for about 1800s [4]. How the measured current is evaluated is
shown in chapter 3.2.3.
20
Figure 3.8 Principle of IRC- and RVM-measurement [13].
Technical data for the KDA-1 [27]
• Formation voltage 1 kV DC.
• Cable types PE/XLPE.
• Cable capacity 3 nF − 1.2 μ F or approximately 4 km.
• Testing time 1 hour per phase.
• Nominal voltage range of the cable under test is 6 kV – 36 kV.
• Operating temperature 0 − 40D C .
3.5 Method used to analyse field measurements from 2007
The main part of this thesis work is, as mentioned earlier, based on field measurements made
on medium voltage cables in the Ekerö grid. When analysing measurements, it is important to
know and take into consideration the circumstances around the measurements. In this case the
following conditions have been taken into account when the analysis was made.
•
•
•
•
•
•
•
The type of cables and terminations and if there are any joints in the cable.
If the terminations were disconnected from the isolators or not.
If the terminations were cleaned before the measurement took place.
The air- temperature and humidity.
The average load flow on the cable.
If the cable has been in service before the measurement took place.
The length of the cable and year of installation.
When the measurements were finished the graphs were plotted and analysed. The main
criteria that were analysed were, if the curves were frequency- and/or voltage dependent and
if the graph showed any hysteresis.
3.5.1 Criteria for analysis of IDA 200 HVU measurements
What generally could be said about non-aged cables is following: In XLPE-cables without
water trees the dielectric losses ε ′′ are very low, at the level of 1 : 8 ⋅ 10 −4 . The material
properties has no voltage dependence which means that the material is linear at least up
21
to U 0 and for most non-aged cables linear as high as up to 2U 0 . However, for some cables
there could be a small non-linear effect detected at 2U 0 . Which means that there is a small
change in the dielectric losses ε ′′ and in the apparent permittivity Δε ′ of less than 3 ⋅10−4 .
Measurements in non-aged cables are always individually reproducible. This means that
neither the voltage pre-stressing of the cable nor the humidity level within the insulation has
any effect on the measurement. XLPE-cables with water trees show a voltage-dependence and
a hysteretic behavior in loss factor ε r′′ and in apparent permittivity Δε r′ . Water tree infested
cables can be separated in three different characteristic responses [2]:
•
VDP Voltage dependent permittivity. VDP means an increase in dielectric losses and
apparent permittivity depending on the voltage level, but almost independent on the
frequency. The dielectric losses and the apparent permittivity shows hysteresis, which
means that the level at the first frequency-sweep is lower than the level of the second
frequency-sweep at the same voltage level. The reason for that is that the water tree
structure in the cable has opened up and leads a larger current [2].
•
TLC Transition to Leakage Currents. TLC is a VDP-response already at the first low
voltage levels. The response changes typically at higher voltage levels [2].
•
LC Leakage Current. This response appears already at low voltage levels and the
hysteretic effect in the dielectric losses is usually larger than for the other two
responses. When the frequency-sweep with the highest voltage has been done, will the
LC-response show higher losses at the same voltage level. One characteristic thing
with the LC response is that the dielectric losses increase with decreasing frequency
with a slope equal to -1 [2].
3.5.2 Analysis of different cable designs
Three different cable designs will be described here, due to their similarity to the cables that
have been measured at the Ekerö grid.
Cable type #A Sievert Kabelverk 24 kV Cable with Extruded Insulation Shield.
This cable has an extruded insulation shield and was manufactured between 1974-1979/80 by
Sievert Kabelverk. This cable type gives a large dielectric response especially at a voltage
higher than the service voltage level. Deteriorated cables often show large bush-like trees,
which are consistently spread around insulation surface. The cable is often not evenly aged,
only parts of the cable could be deteriorate by water trees while other fractions of the cable
could be healthy. This fact makes it difficult to separate a moderately aged from a severely
aged cable. Water tree deteriorated cables shows initially a VDP response and if a high
enough voltage is used they show a TLC response. The required voltage level for detection of
water trees is about 14 kV [2].
Cable type #B Sievert Kabelverk 12 kV Cable with Insulation Shield made of tape only.
This cable has an insulation shield made of tape only and was manufactured between the
years 1966-1974 by Sievert Kabelverk. The water tree of this cable type could be both bushlike and thin-vented trees. The dielectric response of deteriorated cables of this type is
moderate and a VDP response becomes more distinct if the voltage is increased a bit higher
than service voltage level. Usually water tree deteriorated cables are detected by using voltage
levels up to service-voltage level, the voltage breakdown strength is also rather high. When a
22
VDP response has been detected, it usually shows that water trees deteriorate the cable.
However, when the cable type still has a VDP response up to 1.5 times, the voltage servicelevel, the cable can still operate for a few more years [2].
Cable type #C Liljeholmen 24 kV Cable Graphite and Tape Insulation Shield.
This cable has an insulation shield made of graphite paint and tapes and was manufactured by
Liljeholmen in the early 1970s. The water trees of this cable type are usually many thin water
trees. When a cable is deteriorated by water trees the dielectric loss factor is seldom below
3 ⋅10−3 and detectable far below U 0 , the cables voltage dependence is already obvious at
0.25U 0 − 0.5U 0 [2].
If a LC or a TLC response is detected the water tree deterioration in the cable is severe and
the failure risk is very high. If there are no leakage currents detected at a voltage up to U 0 and
the dielectric losses up to U 0 is less than 8 ⋅10−3 then the cable is healthy enough to remain in
service for many years. One characteristic thing with this cable type is that it ages consistently
[2].
Ageing in different cable constructions according to SINTEF [4].
• Local defects are more common in cables with extruded insulation shield.
• Graphite paint and tape insulated cable materials are usually more evenly aged.
3.5.3 Dielectric response in terminations and cable joints
Terminations
There are two types of termination stress cones and material-field grading with different
responses. The influence of terminations is less at higher frequencies [2].
•
•
Stress cones, have a linear response and the field grading is geometrical.
Material field-grading with a stress-grading material has a non-linear response that
starts far below design voltage.
Cable joints
Joints constructed with field-grading material that have changing characteristics, can generate
LC-losses, which dominate the losses from the cable circuit [2].
3.5.4 Temperature dependence of the response
Field measurements show that the temperature has a small influence on water treed XLPEcables with large VDP response. Cables with a large VDP response have changes in dielectric
losses at different temperatures. Nevertheless, the changes are too small to have an influence
on the diagnosis of the cable. A water-treed cable depends also on the humidity within the
insulation, a high response interpret a high humidity level in the insulation. The humidity
level within the cable insulation can decrease if the surrounding temperature rises. Each cable
design has its own influence on the water trees in the cable. In different cable designs there
are different types of water tree structures. One cable model could have only one type of
water tree structures, while another cable model could have several different types of
structures in the same cable [2].
23
3.6 Qualitative dielectric models of water deteriorated cables
To illustrate the complex behaviour of water tree deteriorated insulation, this concept model
shown in figure 3.9 was developed. The model consists of healthy insulation with an area of
deteriorated insulation in form of different kinds of water trees. The arrow shows the increase
direction of the insulation degradation. The resistance drop when the degradation increases.
This means that the resistance that illustrates the background losses is higher than the
resistance that illustrates the LC response. The VDP response part of the model illustrates a
bush tree or a bow-tie tree and the TLC and LC response illustrates a leakage tree. The
difference with TLC and LC are that the LC response occurs already at low voltage while the
TLC response needs a higher voltage to appear.
Increased degradation
AC
Healthy insulation
Background
losses
VDP
response
TLC
response
LC
response
Healthy insulation
Figure 3.9 Concept model of cable insulation.
Three quantitative concept models were developed to express the three types of water trees.
The different water trees are shown in figures 3.10 and 3.11. VDP response is given by bush
trees and bow-tie trees. TLC or LC responses are shown when water trees have grown
through the insulation. The models are shown in the figures 3.13, 3.16 and 3.18.
Leakage tree
Bush tree
Bow-tie tree
Figure 3.10 XLPE-cable deteriorated by
leakage and bush trees [29].
Figure 3.11 XLPE-cable deteriorated by a bow-tie
tree [29].
24
Two plots were made to illustrate the response behaviour of VDP, TLC and LC responses.
The first shown in figure 3.14 illustrates the behaviour of a VDP response. The second plot
showed in figure 3.19 illustrates the behaviour of a TLC or a LC response. The plots are
based on numerical values of one metre long 195 mm2 20 kV cable with 100 trees/metre and
5,5 mm insulation thickness.
The water tree resistance and capacitance for each bush tree that is assumed to be 5 mm wide
and with a length of 4 mm is given by eq. 3.5 and eq. 3.6 and have a cylinder shape. The
response is shown in figure 3.14.
RW = ρ W
dW
d
4 ⋅ 10 −3
= ρW W2 = 1012
A
πr
π 2.5 ⋅ 10 −3
(
)
2
≈ 2 ⋅ 1014 Ω
(3.5)
π ( 2.5 ⋅10
π ( D / 2)
A
CW = ε 0ε r
= ε 0ε r
= 8.854 ⋅10−12 ⋅ 2.3
d − dW
d − dW
1.5 ⋅10−3
Where:
)
−3 2
2
≈ 3 ⋅10−13 F
(3.6)
ρW is the resistivity of the water tree
d W is the length of the water tree
d is the insulation thickness
A is the area of the water tree
D is the width of the water tree
ε 0 is the permittivity in vacuum
ε r is the relative permittivity for a XLPE-cable
The capacitance of the healthy insulation is given by eq. 3.7:
C 0 = length ⋅
2πε 0 ε r
2π ⋅ 8.854 ⋅ 10 −12 ⋅ 2.3
= 1⋅
≈ 2.5 ⋅ 10 −10 F
log(Outerdiametre / Innerdiametre )
log(26.757 / 15.757 )
(3.7)
Where eq. 3.8 gives the inner diameter of a 195 mm2 cable:
Innerdiametre = 2 ⋅
A
π
= 2⋅
195
π
= 15.757 mm
(3.8)
A 195 mm2 got an outer diameter, which is the inner diameter plus the insulation thickness of
11 mm ⇒ 26.8 mm
25
For the leakage tree shown in figure 3.17 has following values have been used for each tree:
The tree is assumed to be 4 mm wide and the length is through the whole insulation of 5.5
mm. This gives:
dW
5.5 ⋅ 10 −3
7
= 10
= 2.5 ⋅ 1010 Ω
RW = ρ W
−3
−3
A
4 ⋅ 10 ⋅ 5.5 ⋅ 10
(3.9)
3.6.1 Equivalent model for bush tree deteriorated cables
Figure 3.12 XLPE-cable deteriorated by
bush tree [28].
Figure 3.13 Equivalent model of a bush tree.
Figure 3.12 above shows an XLPE cable, which is deteriorated by a bush tree. The model for
the bush tree deteriorated cable is shown in figure 3.13 above. To express the dissipation
factor tan δ , the complex capacitance C has to be derived. The capacitance is derived by
expressing the currents through the insulation (outer semi-conductor). This is shown in eq.
3.10 – 3.12. It should be mentioned that the resistance of the water tree RW is voltage
dependent which has not been considered in this simple model.
i = iC0 + i R0 + iW = jωC 0U +
⎛ ω 2 CW2 RW + jωCW
1
U + ⎜⎜
2
R0
⎝ 1 + (ωCW RW )
⎞
⎟U =
⎟
⎠
(3.10)
⎞
⎛
⎟
⎜
2
ωCW RW
CW
1
⎟
⎜
jωU ⎜ C 0 +
+
+
2
2 ⎟
jωR0 j + j (ωCW RW ) 1 + (ωCW RW )
⎟⎟
⎜⎜ ~
C =C ′ − jC ′′
⎠
⎝
26
⎛
⎞
⎜
⎟
⎜
⎟
CW
~
⎜
⎟−
C = C ′ − jC ′′ = ⎜ C 0 +
2 ⎟
1 + (ωC R )
⎟
⎜ WW
C0 >>CW
⎜⎜ ⎟⎟
C′
⎝
⎠
⎛
⎞
⎜
⎟
2
ωCW RW ⎟
1
⎜
+
j
⎜ ωR0 1 + (ωCW RW )2 ⎟
⎜ ⎟
C ′′
⎝
⎠
(3.11)
The term R0 is much larger than RW , however, R0 is neglected because it will not contribute
to the response behaviour of the bush tree plot. The bush tree behaviour is given by the
dissipation factor Δ tan δ Bush−tree :
Δ tan δ Bush −tree =
C ′′
=
C′
ωCW2 RW
2
1 + (ωCW RW )
C0 +
(3.12)
CW
1 + (ωCW RW )
2
Where:
C 0 represents the capacitance of the healthy insulation
R 0 represents the resistance of the healthy insulation
C W represents the capacitance of the healthy insulation in serie with the water tree
R W rep resents the resistance through the water tree
C is the complex capacitance
C ' represents the real part = true capacitance
C '' represents the lossy part
Model of bush tree deteriorated cable
-1
10
-2
10
-3
tandelta
10
-4
10
-5
10
-6
10
-4
10
-3
10
-2
10
-1
10
Frequency Hz
Figure 3.14 Model of bush tree deteriorated cable.
27
0
10
1
10
2
10
Figure 3.14 shows the response from a bush tree deteriorated cable. The cable is assumed to
have 100 water trees/metre. In the low frequency span between 0.1 Hz to 1 Hz where IDA
200 measures, figure 3.14 shows that the dissipation factor is between 3 ⋅ 10 −4 to 4 ⋅ 10 −3 . This
is reasonable values compared to field measurements. However, the values used in the
calculation are idealised and number of water trees is guessed.
3.6.2 Equivalent model for bow-tie tree
Figure 3.15 shows the deterioration of a bow-tie tree below. The model for the bow-tie tree
deteriorated cable, shown in figure 3.16 below, has the same characteristic as the bush tree
model. Except that the resistance is higher and the capacitance is lower than the numerical
values for the bush tree model. This gives the equivalent model shown in figure 3.16 beneath.
The expression for the dissipation factor tan δ shown in eq. 3.15 and eq. 3.16 is derived
exactly as for the bush tree model.
Figure 3.15 XLPE-cable deteriorated by
bow-tie tree [29].
Figure 3.16 Equivalent model of a bow-tie tree.
⎛
⎞
⎜
⎟
⎜
⎟
CW
~
⎜
⎟−
C = C ′ − jC ′′ = ⎜ C 0 +
2 ⎟
1 + (ωC R )
⎟
⎜ WW
C0 >>CW
⎜⎜ ⎟⎟
C′
⎝
⎠
⎛
⎞
⎜
⎟
2
ω
C
R
1
W W
⎟
+
j⎜
⎜ ωR0 1 + (ωCW RW )2 ⎟
⎜ ⎟
C ′′
⎝
⎠
(3.13)
The term R0 is much larger than RW , however, R0 is neglected because it will not contribute
to the response behaviour of a bow-tie tree plot. The bow-tie behaviour is given by the
dissipation factor Δ tan δ Bow−tie −tree :
Δ tan δ Bow−tie −tree =
C ′′
=
C′
ωCW2 RW
2
1 + (ωCW RW )
C0 +
(3.14)
CW
1 + (ωCW RW )
2
28
The response for a bow-tie tree has the same characteristic appearance as for a bush tree, see
figure 3.14.
3.6.3 Equivalent model for a leakage tree
Figure 3.17 shows the deterioration of a leakage tree. The model for the leakage tree
deteriorated cable is shown in figure 3.18 below. The dissipation factor tan δ is derived as for
the bush and bow-tie trees, by expressing the currents through the material eq. 3.15 – 3.17.
Figure 3.17 XLPE-cable deteriorated by a
leakage tree [29].
i = iC 0
Figure 3.18 Equivalent model of a leakage tree.
⎞
⎛
⎟
⎜
1
1
1
1 ⎟
⎜
U = jω U C 0 +
+ i R0 + iW = jωC 0U + U +
+
⎜
R0
RW
jωR0 jωRW ⎟
⎟
⎜ ~
C = C ′ − jC ′′
⎠
⎝
⎛
⎞
⎜ 1
⎟
1
~
⎜
⎟
+
C = C ′ − jC ′′ = C 0 − j
N
ω
R
ω
R
⎜
W ⎟
0
C′
⎜ ⎟
C ′′
⎝
⎠
(3.15)
(3.16)
The term R0 is much larger than RW , however, R0 is neglected because it will not contribute
to the response behaviour of the leakage tree plot. The leakage tree behaviour is given by the
dissipation factor Δ tan δ Leakage−tree :
Δ tan δ Leakage−tree
1
C ′′ ωRW
=
=
C′
C0
(3.17)
29
Model of leakage tree deteriorated cable
1
10
0
10
-1
10
tandelta
-2
10
-3
10
-4
10
-5
10
-6
10
-4
10
-3
10
-2
10
-1
10
Frequency Hz
0
10
1
10
2
10
Figure 3.19 Model of leakage tree deteriorated cable.
Figure 3.19 above shows the characteristic leakage response with the -1 slope.
3.7 Method used to analyse the statistics
For this thesis work statistics were available from Fortum’s Stockholm grid for years 2005
and 2006 [32]. Statistics and failure reports for Ekerö’s medium voltage grid were available
for the years 2004 to 2007 [33]. For the Stockholm grid the total number of faults was
summed up for each year. The number of cable faults was distributed under the categories
high voltage, low voltage and customer. For the high voltage level the damage on the system
is also presented for each year. Moreover, the fault frequency is presented in faults/100 km
cable and year. A small compilation of the consequences of Fortum’s Stockholm grid, caused
by the failures, was also made for year 2005 and 2006. The fault frequency of Stockholm and
Ekerö grid has also been compared to Göteborg Energi and to Darwin. Darwin is a national
database to which most of the energy companies in Sweden report their failure statistics.
The statistics available for Ekerö’s medium voltage grid between 2004 and 2007 were both
from Ekerö Energi’s own statistics and statistics reported to Darwin. The failure reports that
are the basis of the statistics were also available during these years. Even more information
for some of the fault occasions was available through Henrik Svensson. A compilation for
each year was made with all available information. The fault frequency in form of faults/100
km cable and year was also presented.
30
3.8 Validity and reliability
Validity in designation of information is reliable or not. This means that the performed
measurements really gives the answers to the questioned issues, which underlie the
examination [30]. The validity in the measurements made in this thesis work is according to
the author good due to an accurate description of the measurements in the report.
Reliability is a measure on the dependability, which means that same results should be
achieved if measurements are made under the same circumstances [31]. The author considers
the reliability of the measurements as good due to an exact description of the procedure in the
report. This is provided that the statistics are available.
31
4
Results
4.1 Introduction to the chapter
All commercial insulation diagnostic instruments available on the market at this time are all
off-line instruments. The techniques used in most instruments to diagnose water trees in
XLPE-cables are to measure the dissipation factor as a function of frequency. The other
technique used is to diagnose XLPE-cables is to measure the Isothermal Relaxation Current
IRC. The technique used to diagnose PILC-cables is the Return/Recovery Voltage Method
RVM. All instruments are considered to be non-destructive for voltage levels up to U 0 . There
are three manufactures of fault location instruments revealed in this report. All instruments
use the Time Domain Reflectometry TDR.
Cable diagnostic measurements were made on the Ekerö Energi grid on different medium
voltage cables in Stenhamra village, during the period of 03/09/98 – 17/09/98. As part of this
thesis work, measurements have been carried out on various interesting cables, which also
were measured in 1998. The purpose has been to try to determine whether the measurements
made in 1998 were trustworthy. The reliability of the measuring method used in 1998 has
been questioned in a report from the Norwegian company, SINTEF Energiforskning AS [15],
as being not sufficiently reliable.
The measurements in 2007 were made with the IDA 200 HVU system, which measures the
capacitance and the dissipation factor tan δ , explained in chapter 2.2.1. A couple of
measurements were also made with a new instrument under development using TDR, time
domain reflectometry. The instrument is developed by Valentinas Dubickas as a part of his
Ph. D. thesis at KTH, and can be used either for measurements off-line or on-line.
Statistics from Fortum’s Stockholm grid and Ekerö’s medium voltage grid was analysed and
compared with statistics from Göteborg Energi and Darwin.
4.2 SINTEF’s analysis of the measuring instrument used on the
Ekerö grid in 1998
The analysis is based on the dynamics of the relaxation process within the insulation. When
the insulation becomes old or in one way or another becomes worn-out it will bring changes
to the polymer structure. These changes are shown in the relaxation process. When the cable
has been charged, in this case with 1 kV for 1800 seconds, the cable will be discharged for a
short period of time. This discharge of the cable formation creates relaxation currents in the
polymer with different time constants, which are used to evaluate the condition of the cable
insulation [14], [15].
4.2.1 The report of SINTEF Energiforskning AS (SEfAS)
This Nordic project started in the spring in 1997 and was finished at the end of 1999 when the
technical report was published [4]. The aim with the project was to test four commercial cable
diagnostic instruments. In this report only Seba Dynatronic KDA-1 will be of interest. The
diagnoses were made at eight 12 – 24 kV (with insulation thickness 3.4 and 5.5 mm) water
treed XLPE-cables with different constructions. The four chosen cable types were:
32
•
•
•
•
#D a cable with insulation shield made of graphite.
#E a cable with strippable insulation shields.
#F a cable with paint and tape insulation shields.
#G a cable with steam cured insulation shield.
4.2.2 The test procedure
All cables have been installed in the grid. One cable was diagnosed in the field while the
others were diagnosed at SEfAS/NTNU high voltage laboratory in Trondheim Norway. The
diagnoses were made without removing terminations and joints.
When the diagnostic tests were finished all cables were tested with a destructive AC step test.
This test was made to see how many times U 0 the cables actually could resist. This test did
then become the reference when the test instruments should be evaluated.
The tests started with supplying the cable with the voltage U 0 for five minutes, and then
increase the voltage with U 0 every five minutes until breakdown occurred.
After the breakdown test, parts of the cable were sliced into thin sections and examined under
a microscope. In every section the longest vented tree, the longest bow-tie tree and the
number of trees in each section were registered.
4.2.3 Test results
•
•
•
Seba Dynatronic KDA-1 diagnosed cable type #D, with an insulation shield made of
graphite and paint as follows. Phase number one and phase number two were detected
with too high voltage breakdown strength while phase number three was correctly
detected.
Seba Dynatronic KDA-1 diagnosed cable type #E, with strippable insulation as
follows. The detection failed on all three phases. Again the voltage breakdown
strength was detected as too high.
Seba Dynatronic KDA-1 diagnosed cable type #F, with an insulation shield made of
paint and tape correctly.
The test results showed that cable type #D and cable type #E had water trees across the whole
insulation. SEfAS/NTNU showed that the cable type’s #D and #E could be correctly
diagnosed by measuring the polarisation current.
One explanation, that seba did not detect cable type #D and cable type #E correctly, could be
that the instrument only measures the depolarisation current and not both polarisation- and
depolarisation current. The polarisation current is measured during the charging of the cable,
which makes it possible to detect severe water treed deteriorated cables. Seba KDA-1
measures the depolarisation current which is done without any voltage connected to the cable.
That gives a lower current and the cable is detected with too high voltage breakdown strength
[4]. However, even if seba had measured the polarisation current, had probably the formation
voltage of 1 kV been a too low voltage, to make the structure open in cables with water trees.
This means that the cables that are severely deteriorated by leakage trees would probably not
have been detected.
33
It should be mentioned that sebaKMT in their new instrument CDS, measures both
polarisation and depolarisation currents and that the charging voltage is 5 kV [13].
4.3 The analysis of the measurement in 2007
The measurements were limited to three days. The power company Ekerö Energi had two of
their electricians available to support disconnection and reconnection of the cables. It was
judged to have time to measure seven cables during the three days. The chosen cables were
installed between 1970 and 1989 see table 4.1 below.
Table 4.1 XLPE-cables investigated in the study, installed between 1970 and 1989.
Cable number #1
#2
#3
#4
NS2165
NS2123
NS2129
Measurements NS2129
Blåbacksvägen Alviksvägen
Klyvarestigen
Blåbacksvägen
made from
NS2134
NS2160
NS2126
NS2128
End of cable
Klevbersvägen Galtuddsvägen Uppgårdsskolan Skogsvägen
Yes
No
Yes
No
Disconnected
Manufacture
IKO
3x95
Design voltage
Inner semiconductor
24 kV
Extruded
24 kV
Extruded
Outer semiconductor
Extruded
Black tape
Year of
installation
1981
1970
Number of
joints
Terminations
0
0
Red Raychem
Red Raychem
Cable length
A-factor and
classification
in 1998
Classified
breakdown
voltage
strength 1998
650m
1.75, Older
400m
1.99, Older
11
U0
AXKJ
Liljeholmen
AXKJ 3x95
12
U0
#6
NS2132
Lupingränd
NS2131
Humlegränd
Yes
#7
NS2123
Klyvarestigen
NS2122
Strandvägen
Yes
Unknown
AXKJ 3x95
#5
NS2165
Alviksvägen
NS2137
Ramvägen
Yes only at
NS2137
Liljeholmen
AXKJ 3x95
Unknown
AXKJ 3x95
Unknown
AXCEL
3x95, AXCE
3x1x95
24 kV
Extruded
24 kV
Extruded
24 kV
Extruded
24 kV
Extruded
Extruded
Black tape
Black tape
Extruded
1980
1971
1971
1986-89
2
0
0
1 Raychem
Red Raychem
at
NS2123,
Elastimould at
NS 2126
Unknown
Unknown,
Critical
Red Raychem
Black tape
White tape
Red Raychem
300m
2.44, Critical
600m
2.7, Critical
250m
2.36,
Critical
450m
2.57, Critical
Unknown
8
Unknown
AXCE
3x1x150, AXKJ
3x95, AXCEL
95
24 kV
Extruded,
Unknown,
Extruded
Extruded,
Unknown,
Extruded
Unknown
AXKJ
from
1971
2
U0
3
U0
4
U0
8
The chosen cables were a mix of cables with and without joints and from the early 70s to the
mid 80s. Before the measurements started the terminations were cleaned with isopropanol. In
these cases when the terminations were not disconnected also the isolators were cleaned with
isopropanol, to avoid creep currents. The average operation current according to Henrik
Svensson at Ekerö Energi is about 50-60 A.
34
U0
4.3.1 Measurements day one
The first day of measurements the temperature was +2D C and the humidity 100% RH. It
started to rain a little in the afternoon. The first cable to measure was cable #5 between
NS2137 Ramvägen and NS2165 Alviksvägen. This is a 24 kV cable manufactured by
Liljeholmen 1971, it is 600 metres long with black tape terminations and without any joints.
The cable was disconnected at Alviksvägen but still connected at 20 cm high red plastic
isolators at Ramvägen. The cable had been out of service for about 11 months, according to
Henrik Svensson at Ekerö Energi. The first phase had burned off directly after the
termination. The measurement took place at Alviksvägen NS2165.
Analysis of cable #5
When the measurements were made in 1998 cable #5 got an A-factor of 2.7 and was
classified as critical. The voltage breakdown strength was estimated to 3 U 0 .
The first phase L1 of cable #5, shown in figures 4.1 and 4.2. Figure 4.1 shows voltage
dependence already at low voltage levels. The losses are frequency dependent, which is
characteristic for both TLC and LC response. However in this case, an LC response is shown
with a slope close to -1 in the lower frequency spectra between 0.1-0.3 Hz. The measurement
also shows a hysteresis, which is especially large between the 3 kV levels. In figure 4.2 the
capacitance as a function of frequency is plotted. Figure 4.2 shows a large VDP response,
which indicates water treeing. This phase is aged and deteriorated by water trees, nevertheless
not grown through the entire insulation. The fact that this phase has burned off after the
termination would probably not have had any impact on the measurement. This phase was
also measured with the time domain reflectometry TDR in off-line mode.
35
Cable #5 Phase L1
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.1 Cable #5 phase L1, tangents delta as a function of frequency.
-7
1.202
x 10
Cable #5 Phase L1
3kVrms
6kVrms
9kVrms
1.201
12kVrms
6kVrms
c´
1.2
3kVrms
1.199
1.198
1.197
1.196
-1
10
0
10
Frequency(Hz)
Figure 4.2 Cable #5 phase L1, the capacitance as a function of frequency.
36
1
10
The second phase L2 of cable #5, shown in figures 4.3 and 4.4 below. Figure 4.3 shows that
this phase has a VDP response and when the frequency decreases the losses and the hysteresis
increase. The capacitance in figure 4.4 shows small VDP response. This phase is aged and
probably deteriorated by small water trees, due to the moderate losses.
Cable #5 Phase L2
-2
10
3kV rms
6kV rms
9kV rms
12kV rms
6kV rms
tan(δ)
3kV rms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.3 Cable #5 phase L2, tangents delta as a function of frequency.
-7
1.195
x 10
Cable #5 Phase L2
3kVrms
6kVrms
1.1945
9kVrms
12kVrms
6kVrms
1.194
3kVrms
c´
1.1935
1.193
1.1925
1.192
1.1915
-1
10
0
10
Frequency(Hz)
Figure 4.4 Cable #5 phase L2, the capacitance as a function of frequency.
37
1
10
The third phase L3 of cable #5, shown in figures 4.5 and 4.6 below. Figures 4.5 and
4.6 shows small VDP response. The losses are larger than for phase L2 but lower than
for phase L1. This phase is aged and probably deteriorated by water trees.
Cable #5 Phase L3
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.5 Cable #5 phase L3, tangents delta as a function of frequency.
-7
1.1995
x 10
Cable #5 Phase L3
3kVrms
6kVrms
9kVrms
1.199
12kVrms
6kVrms
3kVrms
c´
1.1985
1.198
1.1975
1.197
-1
10
0
10
Frequency(Hz)
Figure 4.6 Cable #5 phase L3, the capacitance as a function of frequency.
38
1
10
Summary of cable #5
The measurements in 1998 showed that cable #5 got an A-factor of 2.7 and was classified as
critical with estimated voltage breakdown strength of 3 U 0 .
The cable is aged and probably deteriorated by water trees. Phase L1 is more deteriorated by
water trees than phase L2 and L3 where the losses are more moderate. However, the cable
will most likely remain in service for many years.
The second cable to measure was cable #2 between NS2165 Alviksvägen and NS2160
Galtuddsvägen. This is a 24 kV cable manufactured by Liljeholmen in 1970, it is 400 m long
with red Raychem terminations and without any joints. The cable was connected at both ends.
The cable had been in service until a couple of hours before the measurements took place at
Alviksvägen.
Analysis of cable #2
When the measurements were made in 1998 cable #2 got an A-factor of 1.99 and was
classified as older. The voltage breakdown strength was estimated to 12 U 0 .
When the measuring data has been analysed for cable #2, it has been compared to the
behaviour of cable type #B. The characteristic response of a deteriorated cable of this type is
moderate even though the VDP response becomes more distinct when the voltage increases
over service voltage level. However, the measurements on this cable were only made up to
service-voltage level.
The cable was cleaned a couple of hours before the measurements took place, which might
have destroyed these measurements. Due to the high humidity, damp can have gathered on the
isolators and terminations and caused creep currents.
39
The first phase L1 of cable #2 is shown in figures 4.7 and 4.8 below. The graph in figure 4.7
shows small VDP response and a small hysteresis and also an LC response from 1 Hz and
below. The response indicates a leakage current which is most likely caused by creep
currents, due to the high humidity and that the terminations were not disconnected from the
isolators. The capacitance in figure 4.8 shows no voltage dependence, which indicates that
this phase is in good condition. Phase L2 was not possible to download from the IDA 200
HVU instrument.
Cable #2 Phase L1
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.7 Cable #2 phase L1, tangents delta as a function of frequency.
-7
1.124
x 10
Cable #2 Phase L1
3kVrms
6kVrms
1.1238
9kVrms
12kVrms
6kVrms
1.1236
3kVrms
c´
1.1234
1.1232
1.123
1.1228
1.1226
-1
10
0
10
Frequency(Hz)
Figure 4.8 Cable #2 phase L1, the capacitance as a function of frequency.
40
1
10
The third phase L3, is shown in figures 4.9 and 4.10 below. Figure 4.9 show that this phase
also has small VDP response and almost no hysteresis. The response indicates that creep
currents probably influence the measurements of this phase. The capacitance plot in figure
4.10 shows no voltage dependence. The condition of this phase is, due to no voltage
dependence in the capacitance plot, also probably good.
Cable #2 Phase L3
-2
10
3kVrms
6kVrms
9kVrms
-3
10
12kVrms
6kVrms
tan(δ)
3kVrms
-4
10
-5
10
-6
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.9 Cable #2 phase L3, tangents delta as a function of frequency.
-7
1.0958
x 10
Cable #2 Phase L3
3kVrms
6kVrms
1.0957
9kVrms
1.0956
12kVrms
6kVrms
1.0955
3kVrms
c´
1.0954
1.0953
1.0952
1.0951
1.095
1.0949
-1
10
0
10
Frequency(Hz)
Figure 4.10 Cable #2 phase L3, the capacitance as a function of frequency.
41
1
10
Summary of cable #2
In the measurements in 1998 cable #2 got an A-factor of 1.99 and was classified as older with
estimated voltage breakdown strength of 12 U 0 .
The analysis of this cable is uncertain. Due to the fact that the terminations were not
disconnected from the isolators, and that the cable terminations and the isolators were cleaned
a couple of hours before the measurements took place. Due to the high humidity this day,
damp could have gathered on the terminations and isolators and have had an impact on the
measurements in form of leakage currents. This could be the reason for the LC response on
phase L1 and phase L3. However, the capacitance plots shows no voltage dependence, which
indicates that the cable is in good condition. This cable will probably remain in service for
many years.
4.3.2 Measurements day two
The weather the second day of measuring was similar to the first day, the temperature was
+2D C and the humidity was 100% RH. The first cable to measure was cable #4 between
NS2129 Blåbacksvägen and NS2128 Skogsvägen. The measurements took place at
Blåbacksvägen. This is a 24 kV cable installed in 1980 by an unknown manufacturer. The
cable is 300 metres long with red Raychem terminations and with two joints. The cable was
connected to the isolators at both ends. The cable had been in service just before the
measurements took place.
Analysis of cable #4
When the measurements were made in 1998 cable #4 got an A-factor of 2.44 and was
classified as critical. The voltage breakdown strength was estimated to 8 U 0 .
The manufacture of cable #4 is unknown. However, it was installed in 1980 and cable #1,
which was manufactured by IKO, was installed in 1981 and the both cables are of the same
type. Therefore the analysis of cable #4 has been compared to the behaviour of cable type #A.
This cable type gives a large dielectric response especially at a slightly larger voltage than the
service voltage level. Deteriorated cables often show large bush-like trees. This cable is often
partly deteriorated by water trees while other fractions of the cable could be healthy. Water
tree deteriorated cables initially show a VDP response and if a high enough voltage is used
they show a TLC response. However, the measurements on this cable were only made up to
service-voltage level.
42
The first phase L1 of cable #4 is shown in figures 4.11 and 4.12. Figure 4.11 shows a clear
LC response. Corona discharges were heard already from 9 kV from phase L1 and from 6 kV
from phase L2. All three phases showed the same LC response. The capacitance in figure 4.12
shows no voltage dependence, which indicates that the cable is in good condition. The phases
L2 and L3 were not able to download from the IDA 200 HVU system. The LC response
appears to be most likely due to creep currents over the isolators, seeing comments on cable
#1 below. The cable was also measured with the TDR system, an area with joints was
detected but no sign of ageing was found.
Cable #4 Phase L1
0
10
3kVrms
6kVrms
9kVrms
-1
tan(δ)
10
-2
10
-3
10
-1
10
0
1
10
Frequency(Hz)
10
Figure 4.11 Cable #4 phase L1, tangents delta as a function of frequency.
Cable #4 Phase L1
-7.3017
10
3kVrms
6kVrms
-7.3019
10
9kVrms
-7.3021
c´
10
-7.3023
10
-7.3025
10
-7.3027
10
-1
10
0
10
Frequency(Hz)
Figure 4.12 Cable #4 phase L1, the capacitance as a function of frequency.
43
1
10
Summary of cable #4
In the measurements in 1998 cable #4 got an A-factor of 2.44 and was classified as critical
with estimated voltage breakdown strength of 8 U 0 .
The analysis of this cable is difficult to do, due to the cable was not disconnected from the
isolators. The high humidity this day most likely caused creep currents over the isolators,
which failed these measurements. These creep currents give rises to a LC response. However,
the capacitance in figure 4.12 shows no voltage dependence, which indicates that the cable is
in good condition, and will remain in service in many years. TDR measurements showed the
joints, but no ageing of the cable was found.
The second cable to measure was cable #1 between NS2134 Klevbergsvägen and NS2129
Blåbacksvägen. The measurements took place at Blåbacksvägen. This is a 24 kV cable
manufactured in 1981 by IKO. The cable is 650 metres long with red Raychem terminations
and without any joints. The cable had been in service until the measurements took place. The
cable was not disconnected at first and the response exhibited a leakage current just as for
cable #4. The measurement was stopped and the cable was disconnected at both ends. The
analysis of cable #1 is made with both ends of the cable disconnected.
Analysis of cable #1
When the measurements were made in 1998 cable #1 got an A-factor of 1.75 and was
classified as older. The voltage breakdown strength was estimated to 11 U 0 .
When the measuring data has been analysed for cable #1, it has been compared to the
behaviour of cable type #A. This cable type gives a large dielectric response especially at a
slightly larger voltage than the service voltage level. Deteriorated cables often show large
bush-like trees. This cable is often partly deteriorated by water trees while other fractions of
the cable could be healthy. Water tree deteriorated cables initially shows a VDP response and
if a high enough voltage is used they show a TLC response [2]. However, the measurements
on this cable were only made up to service-voltage level.
44
Phase L1 of cable #1 is showed in figures 4.13 and 4.14 below. Figure 4.13 shows no VDP
response and very low losses. The response is frequency independent except for a small
increase of losses at the highest voltage levels, the graph also shows a very small hysteresis. A
small voltage dependency is shown in the capacitance plot in figure 4.14 which probably only
is measurement insecurity. This phase is in good condition.
Cable #1 Phase L1
-3
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-4
10
-5
10
-1
10
0
1
10
Frequency(Hz)
10
Figure 4.13 Cable #1 phase L1, tangents delta as a function of frequency.
-7
1.199
x 10
Cable #1 Phase L1
3kVrms
6kVrms
1.1988
9kVrms
12kVrms
1.1986
6kVrms
3kVrms
c´
1.1984
1.1982
1.198
1.1978
1.1976
-1
10
0
10
Frequency(Hz)
Figure 4.14 Cable #1 phase L1, the capacitance as a function of frequency.
45
1
10
The second phase L2 is shown in figures 4.15 and 4.16 below. The graph in figure
4.15 shows no VDP response and low losses. The response is almost frequency
voltage independent and there is no hysteresis. Figure 4.16 shows that the capacitance
has a small voltage dependence which probably only is measurement insecurity. This
phase is in good condition.
Cable #1 Phase L2
-3
10
3kVrms
6kVrms
9kVrms
12kVrms
-4
10
6kVrms
tan(δ)
3kVrms
-5
10
-6
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.15 Cable #1 phase L2, tangents delta as a function of frequency.
-7
1.196
x 10
Cable #1 Phase L2
3kVrms
1.1958
6kVrms
9kVrms
1.1956
12kVrms
1.1954
6kVrms
3kVrms
c´
1.1952
1.195
1.1948
1.1946
1.1944
1.1942
-1
10
0
10
Frequency(Hz)
Figure 4.16 Cable #1 phase L2, the capacitance as a function of frequency.
46
1
10
The third phase L3 is shown in figures 4.17 and 4.18 below. The graph in figure 4.17 also
shows low losses and no VDP response. The response is almost frequency voltage
independent and there is nearly no hysteresis. The capacitance in figure 4.18 shows a small
voltage dependence which probably only is measurement insecurity. This phase is also in
good condition.
Cable #1 Phase L3
-3
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-4
10
-5
10
-1
10
0
1
10
Frequency(Hz)
10
Figure 4.17 Cable #1 phase L3, tangents delta as a function of frequency.
-7
1.1948
x 10
Cable #1 Phase L3
3kVrms
6kVrms
1.1946
9kVrms
12kVrms
1.1944
6kVrms
3kVrms
c´
1.1942
1.194
1.1938
1.1936
1.1934
-1
10
0
10
Frequency(Hz)
Figure 4.18 Cable #1 phase L3, the capacitance as a function of frequency.
47
1
10
Summary of cable #1
In the measurements in 1998 cable #1 got an A-factor of 1.75 and was classified as older with
estimated voltage breakdown strength of 11 U 0 .
The response of this cable has almost no voltage dependence and no hysteresis and the losses
are very low. The capacitance plots shows a small voltage dependence which probably only is
measurement insecurity. However, this cable is in a good condition and will probably remain
in service for many years. The cable was also measured by the TDR instrument, which did not
show any sign of ageing. Cable #1 was installed one year after cable #4 and is of the same
type of cable. Before disconnecting the terminations, cable #1 had the same behaviour in form
of leakage currents as cable #4 had. It is most likely that cable #4 would have had the same
response as cable #1 if it had been disconnected from the isolators. This means that cable #4
probably also is in a good condition, see chapter 4.3.2.
The last cable measured the second day was cable #7 between NS2122 Strandvägen and
NS2123 Klyvarestigen. The measurements took place at Klyvarestigen. Between these
stations there are two different cables, from Klyvarestigen there are single phase cables about
100 metres, according to Mikael Olofsson at Ekerö Energi, there is a joint which connects a
three phase cable. Both cables are 24 kV cables. The single-phase cables are probably
manufactured in 1988-1989 since the station was built in June 1989, the manufacturer is
although unknown. The three-phase cable was manufactured in 1986 by an unknown
manufacturer. The cable is 450 metres long with red Raychem terminations and with one
Raychem joint. The cable had been in service until the measurements took place. The cable
was disconnected at both ends.
Analysis of cable #7
When the measurements were made in 1998 cable #7 got an A-factor of 2.57 and was
classified as critical. The voltage breakdown strength was estimated to 8 U 0 .
There is no data available on how to analyse the single-phase cables. Due to most of the cable
length being a three phase cable with an extruded insulation shield, the measuring data for
cable #7 has been compared to the behaviour of cable type #A. This cable type gives a large
dielectric response, especially at a slightly larger voltage than the service voltage level.
Deteriorated cables often show large bush-like trees. This cable is often partly deteriorated by
water trees while other fractions of the cable could be healthy. Water tree deteriorated cables,
show initially a VDP response and if a high enough voltage is used they show a TLC response
[2]. However, the measurements on this cable were only made up to service-voltage level.
48
Phase L1 of cable #7 is showed in figures 4.19 and 4.20 below. Figure 4.19 show that this
phase has rather high losses but no VDP response, except for the 12 kV level, and no
hysteresis. The joint could have influenced this measurement and given these high losses. The
capacitance in figure 4.20 shows no voltage dependence, which indicates that this phase is not
deteriorated by any water trees.
Cable #7 Phase L1
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.19 Cable #7 phase L1, tangents delta as a function of frequency.
-7
1.1292
x 10
Cable #7 Phase L1
3kVrms
6kVrms
1.129
9kVrms
12kVrms
1.1288
6kVrms
3kVrms
c´
1.1286
1.1284
1.1282
1.128
1.1278
1.1276
-1
10
0
10
Frequency(Hz)
Figure 4.20 Cable #7 phase L1, the capacitance as a function of frequency.
49
1
10
Phase L2 of cable #7 is shown in figures 4.21 and 4.22 below. Figure 4.21 shows
almost the same characteristics as phase L1, but lower losses, no VDP response except
for 9 kV and 12 kV and no hysteresis. The capacitance in figure 4.22 shows no voltage
dependence, which indicates that this phase probably not is deteriorated by any water
trees.
.
Cable #7 Phase L2
-2
10
3kV rms
6kV rms
9kV rms
12kV rms
6kV rms
tan(δ)
3kV rms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.21 Cable #7 phase L2, tangents delta as a function of frequency.
-7
1.133
x 10
Cable #7 Phase L2
3kVrms
1.1328
6kVrms
9kVrms
1.1326
12kVrms
6kVrms
1.1324
3kVrms
c´
1.1322
1.132
1.1318
1.1316
1.1314
1.1312
1.131
-1
10
0
10
Frequency(Hz)
Figure 4.22 Cable #7 phase L2, the capacitance as a function of frequency.
50
1
10
Phase L3 of cable #7 is shown in figures 4.23 and 4.24 below. The third phase shown
in figure 4.23 has also rather high losses, which probably is an effect of the joint. No
hysteresis and no VDP response are shown at all. The capacitance in figure 4.24
shows no voltage dependence. This phase also looks good and is probably not
deteriorated by any water trees.
Cable #7 phase L3
-2
10
3kV rms
6kV rms
9kV rms
12kV rms
6kV rms
tan(δ)
3kV rms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.23 Cable #7 phase L3, tangents delta as a function of frequency.
-7
1.133
x 10
Cable #7 Phase L3
3kVrms
1.1328
6kVrms
9kVrms
1.1326
12kVrms
6kVrms
1.1324
3kVrms
c´
1.1322
1.132
1.1318
1.1316
1.1314
1.1312
1.131
-1
10
0
10
Frequency(Hz)
Figure 4.24 Cable #7 L3, the capacitance as a function of frequency.
51
1
10
Summary of cable #7
In the measurements in 1998 cable #7 got an A-factor of 2.57 and was classified as critical
with estimated voltage breakdown strength of 8 U 0 .
The response of this cable has no voltage dependence and no hysteresis. The losses are rather
high probably due to the joint. The capacitance shows no voltage dependence. This indicates
that the cable is in good condition and probably not deteriorated by water trees. The cable will
most likely remain in service for many years.
4.3.3 Measurements day three
On the last day of measurements, the temperature had increased to +8D C and the humidity was
100% RH. The first cable to measure was cable #6 between NS2132 Lupingränd and NS2131
Humlegränd. The measurements took place at Lupingränd. This is a 24 kV cable
manufactured in 1971. The manufacturer is probably Liljeholmen, due to the fact that other
cables from the early 70s used in the Ekerö grid were manufactured by Liljeholmen.
According to Mikael Olofsson at Ekerö Energi they usually used the same manufacturer at
this time. The cable is 250 metres long with white tape terminations and without any joints.
The cable was disconnected from the isolators at both ends. The cable had been in service just
before the measurements took place.
Analysis of cable #6
When the measurements were made in 1998 cable #6 got an A-factor of 2.36 and was
classified as critical. The breakdown voltage was estimated to 4 U 0 .
When the measuring data has been analysed for cable #6, it has been compared to the
behaviour of cable type #B. The characteristic response of a deteriorated cable of this type is
moderate even though the VDP response becomes more distinct when the voltage increases
over service voltage level [2]. However, the measurements on this cable were only made up to
service-voltage level.
52
Phase L1 of cable #6 is shown in figures 4.25 and 4.26 below. The graph of phase L1 shows a
VDP response but no hysteresis and the losses are moderate. The capacitance in figure 4.26
shows small voltage dependence. This phase is aged and probably deteriorated by small water
trees. However, due to the moderate losses indicate that the ageing of this phase is moderate
and not critical.
Cable #6 Phase L1
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
10
-4
10
-1
0
10
10
Frequency(Hz)
1
10
Figure 4.25 Cable #6 phase L1, tangents delta as a function of frequency.
-8
6.728
x 10
Cable #6 Phase L1
3kVrms
6kVrms
6.726
9kVrms
12kVrms
6kVrms
6.724
3kVrms
6.722
c´
tan(δ)
3kVrms
-3
6.72
6.718
6.716
6.714
-1
10
0
10
Frequency(Hz)
Figure 4.26 Cable #6 phase L1, the capacitance as a function of frequency.
53
1
10
The second phase L2 of cable #6 is shown in figures 4.27 and 4.28 below. The characteristic
shown in figure 4.27 is similar to phase L1, a small VDP response but no hysteresis and
moderate losses. The capacitance in figure 4.28 shows almost no voltage dependence. This
phase is aged and probably deteriorated by small water trees. However, due to the moderate
losses indicate that the ageing of this phase is moderate and not critical.
Cable #6 Phase L2
-2
10
3kV rms
6kV rms
9kV rms
12kV rms
6kV rms
tan(δ)
3kV rms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.27 Cable #6 phase L2, tangents delta as a function of frequency.
-8
6.816
x 10
Cable #6 Phase L2
3kVrms
6kVrms
6.814
9kVrms
12kVrms
6kVrms
6.812
3kVrms
c´
6.81
6.808
6.806
6.804
6.802
-1
10
0
10
Frequency(Hz)
Figure 4.28 cable #6 phase L2, the capacitance as a function of frequency.
54
1
10
The third phase L3 of cable #6 is shown in figures 4.29 and 4.30 below. This phase has the
same behaviour as phases L1 and L2. Figure 4.29 shows a small VDP response with no
hysteresis and slightly lower losses than L1 and L2 is shown. The capacitance in figure 4.30
shows no voltage dependence. This phase is aged and probably deteriorated by small water
trees. However, due to the moderate losses indicate that the ageing of this phase is moderate
and not critical.
Cable #6 phase L3
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.29 Cable #6 phase L3, tangents delta as a function of frequency.
-8
6.762
x 10
Cable #6 Phase L3
3kVrms
6kVrms
6.76
9kVrms
12kVrms
6kVrms
6.758
3kVrms
c´
6.756
6.754
6.752
6.75
6.748
-1
10
0
10
Frequency(Hz)
Figure 4.30 Cable #6 phase L3, the capacitance as a function of frequency.
55
1
10
Summary of cable #6
In the measurements in 1998 cable #6 got an A-factor of 2.36 and was classified as critical
with estimated voltage breakdown strength of 4 U 0 .
This cable shows VDP response but no hysteresis in the tangents delta plot. The capacitance
plot shows almost no voltage dependence. The VDP response indicates that small water trees
deteriorate this cable. This is based on the fact that the tangents delta plot does not show any
hysteresis. When the voltage level increases in the cable the water tree structure in the cable
will open up. If the cable had been deteriorated by a lot of large water trees it would probably
have shown in a hysteresis. The cable will probably remain in service for many years. The
cable was also measured with the TDR instrument, the response did not show any ageing
characteristics.
The last cable to measure the last day was cable #3 between NS2126 Uppgårdsskolan and
NS2123 Klyvarestigen. The measurements took place at Klyvarestigen. Between these
stations there are three different cables. There are two joints connecting the cables. All cables
are 24 kV cables. The installation year is unknown. However, the station is probably built at
the end of the 80s, due to the fact that the station on Klyvarestigen was built in 1989 and the
station at Uppgårdsskolan was of a similar type. However, the AXKJ cable in the middle is
from 1971. The manufacturer of the cables and the length of the cables are unknown. At
NS2123 there were red Raychem terminations and at NS2126 there were Elastimoud
terminations. The cable had been in service until the measurements took place. The cable was
disconnected at both ends. TDR measurement was made on phase 1. The TDR measurement
did not show any ageing.
Analysis of cable #3
When the measurements were made in 1998 cable #3 was not classified with either A-factor
or breakdown voltage strength. However, it was classified as critical.
56
Phase L2 of cable #3 is shown in figures 4.31 and 4.32 below. The graph in figure 4.31 shows
a voltage independent leakage current with no hysteresis. The capacitance in figure 4.32
shows no voltage dependence, which indicates that there is no water tree deterioration. The
cable has two joints, which could be the reason for this response. This phase is probably in a
fairly good condition. The response for phase L1 was not able to download from the IDA 200
HVU system, however, it showed the same response as phase L2.
Cable #3 Phase L2
-2
10
3kVrms
6kVrms
9kVrms
12kVrms
6kVrms
tan(δ)
3kVrms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.31 Cable #3 phase L2, tangents delta as a function of frequency.
-7
1.587
x 10
Cable #3 phase L2
3kVrms
1.5865
6kVrms
9kVrms
1.586
12kVrms
6kVrms
1.5855
3kVrms
c´
1.585
1.5845
1.584
1.5835
1.583
1.5825
1.582
-1
10
0
10
Frequency(Hz)
Figure 4.32 Cable #3 phase L2, the capacitance as a function of frequency.
57
1
10
Phase L3 is shown in figures 4.33 and 4.34 below. Figure 4.33 shows a leakage
current and small voltage dependence at 9 kV and 12 kV but no hysteresis. The cable
joints also probably affect this response. The capacitance in figure 4.34 shows no
voltage dependence. This phase is most likely not deteriorated by water trees and in
good condition.
Cable #3 Phase L3
-2
10
3kV rms
6kV rms
9kV rms
12kV rms
6kV rms
tan(δ)
3kV rms
-3
10
-4
10
-1
0
10
1
10
Frequency(Hz)
10
Figure 4.33 Cable #3 phase L3, tangents delta as a function of frequency.
-7
1.589
x 10
Cable #3 Phase L3
3kVrms
1.5885
6kVrms
9kVrms
1.588
12kVrms
6kVrms
1.5875
3kVrms
c´
1.587
1.5865
1.586
1.5855
1.585
1.5845
1.584
-1
10
0
10
Frequency(Hz)
1
10
Figure 4.34 Cable #3 phase L3, the capacitance as a function of frequency.
Summary of cable #3
In the measurements in 1998 cable #3 was classified as critical but with no A-factor and no
estimated breakdown voltage strength. This cable is difficult to analyse due to the cable
consists of three different parts. All three phases in this cable show a voltage independent
58
leakage current response with no hysteresis. The capacitance did not show any voltage
dependence either. This fact indicates that the cable is in a good condition and not
deteriorated by water trees. The leakage current response is most likely an effect of the two
cable joints. This cable will probably remain in service in many years.
4.3.4 Results of the measured XLPE-cables made in 2007
Table 4.2 below summaries all information available on the measured cables. It also shows
the results from the measurements made in 1998 and in 2007. Table 4.2 also contains a
suggestion for re-measurements, which is based on: year of installation and cable type and the
results of the analysis made in 2007.
Table 4.2 Results of the XLPE-cables investigated in the study, installed between 1970 and 1989.
Cable number #1
#2
#3
#4
#5
NS2165
NS2123
NS2129
NS2165
Measurements NS2129
Blåbacksvägen Alviksvägen
Klyvarestigen
Blåbacksvägen Alviksvägen
made from
NS2134
NS2160
NS2126
NS2128
NS2137
End of cable
Klevbersvägen Galtuddsvägen Uppgårdsskolan Skogsvägen
Ramvägen
Yes
No
Yes
No
Yes only at
Disconnected
NS2137
IKO
AXKJ Liljeholmen
Unknown
Unknown
Liljeholmen
Manufacture
3x95
AXKJ 3x95
AXCE
AXKJ 3x95
AXKJ 3x95
3x1x150, AXKJ
3x95, AXCEL
95
24 kV
24 kV
24 kV
24 kV
Design voltage 24 kV
Extruded
Extruded
Extruded,
Extruded
Extruded
Inner semiUnknown,
conductor
Extruded
Extruded
Black tape
Extruded,
Extruded
Black tape
Outer semiUnknown,
conductor
Extruded
1981
1970
Unknown
1980
1971
Year of
AXKJ
from
installation
1971
0
0
2
2
0
Number of
joints
Red Raychem
Red Raychem
Red Raychem Red Raychem
Black tape
Terminations
at
NS2123,
Elastimodul at
NS 2126
650m
400m
Unknown
300m
600m
Cable length
1.75, Older
1.99, Older
Unknown,
2.44, Critical
2.7, Critical
A-factor and
Critical
classification
in 1998
Unknown
Classified
12 U 0
8 U0
3 U0
11 U 0
breakdown
voltage
strength 1998
The cable is in The cable is
Analysis made The cable is in The cable is in Difficult to
good condition good condition analyse due to
good condition aged and
in 2007
and probably
and probably three different
and probably probably
not
not
cables.
not
deteriorated
deteriorated by deteriorated by However, the
deteriorated by by large
water trees.
water trees.
cable is
water trees.
water trees.
probably in
good condition
and not
deteriorated by
water trees.
Within 10
Within 10
Within 5-10
Within 10
Within 5
Suggested reyears
years
years
years
years
measuring
59
#6
NS2132
Lupingränd
NS2131
Humlegränd
Yes
#7
NS2123
Klyvarestigen
NS2122
Strandvägen
Yes
Unknown
AXKJ 3x95
Unknown
AXCEL
3x95, AXCE
3x1x95
24 kV
Extruded
24 kV
Extruded
Black tape
Extruded
1971
1986-89
0
1 Raychem
White tape
Red Raychem
250m
2.36,
Critical
450m
2.57, Critical
4
U0
8
U0
The cable is
aged and
probably
deteriorated
by small
water trees.
The cable is
in good
condition and
probably not
deteriorated
by water
trees.
Within 5
years
Within 10
years
4.4 Statistics from the Stockholm grid 2005 and 2006
Failure statistics from Fortum’s Stockholm’s grid were available for 2005 and 2006 [32]. The
failure statistics were available for the high voltage level, the low voltage level and for service
level. A fundamental model of a distribution grid is showed in figure 4.35. The table 4.3
describes the total number of faults and how many of them that were underground cable
faults. The underground cable faults are also presented in percent. The table also shows the
number of faults and percent of totally underground cable faults at each voltage level.
Customer is the customers service cable which could be either high or low voltage. The
Stockholm grid had totally 918 underground cable faults in 2005. By these were 97
high/medium voltage cables faults, which are about 11 percent of the total cable faults. The
remaining 821 were low voltage cables faults or customer cable faults. In 2006 there was a
small increase in total number of cable faults to 938. However, seen to the total number of
faults the cable faults had increased with 20 percent since 2005. By the total number of cable
faults in 2006 109 were high voltage faults, which are about 12 percent of the total cable
faults. The remaining 829 were low voltage cables faults or customer cable faults. Worth
noting is that the low voltage faults has increased with almost 77 percent from 2005 to 2006.
However, two years is a too short period of time to determine if this is a trend or just a normal
statistic variation.
Factory
Service cable
High voltage cable
Public house
24 kV / 6 kV
Feder
High voltage cable
Service cable
Low voltage cable
Building 1
24 kV / 0.4 kV
Service cable
House
Figure 4.35 Describes a fundamental model of a distribution grid.
60
Table 4.3
Year Total
number
of faults
2005
2006
1550
1184
Total
number
of cable
faults
918
938
Total
cable
faults in
% of
total
number
of faults
59,2
79,2
High
voltage
cable
faults
In %
of
total
cable
faults
Low
voltage
cable
faults
In %
of
total
cable
faults
Customer
cable
faults
(Fault at
service
cables)
In %
of
total
cable
faults
97
109
10,5
11,6
108
191
11,8
20,4
713
638
77,7
68,0
The subsequent table 4.4 describes the cause of the high voltage faults. The table also shows
the number of faults at each cause category and how many percent that represents of the total
high voltage faults.
Table 4.4 High voltage underground cables
Cause of fault on
high voltage cables
Lack of maintenance
or worn out cables
Unknown
Unselective
Digging
Incorrect connection
Corrosion
Unselective-region
Sabotage
Overload
Fabrication fault
Lightning
Total
Year
2005
(pcs.)
63
Percent of total high
voltage cable faults
2005 (%)
65,0
Year
2006
(pcs.)
50
Percent of total high
voltage cable faults
2006 (%)
45,9
25
0
6
0
0
1
0
0
1
1
97
25,8
0
6,2
0
0
1
0
0
1
1
100
30
12
11
2
1
1
1
1
0
0
109
27,5
11,0
10,1
1,8
0,9
0,9
0,9
0,9
0
0
99,9
The table 4.5 below describes the cause of the low voltage faults. The table also shows the
number of faults at each cause category and how many percent that represents of the total low
voltage faults.
Table 4.5 Low voltage underground cables
Cause of fault on low
voltage cables
Lack of maintenance
or worn out cables
Unknown
Unselective
Digging
Incorrect connection
Corrosion
Sabotage
Overload
Fabrication fault
Own fuse
Circuit breaker
Year
2005
(pcs.)
48
Percent of total low
voltage cable faults
2005 (%)
44,4
Year
2006
(pcs.)
77
Percent of total low
voltage cable faults
2006 (%)
40,3
19
2
9
0
0
4
11
2
3
1
17,6
1,9
8,3
0
0
3,7
10,2
1,9
2,8
0,9
37
2
18
3
1
2
29
4
0
0
19,4
1,0
9,4
1,6
0,5
1,0
15,2
2,1
0
0
61
Customer fault
Montage fault
Region grid
Traffic
Animals
Total
1
2
1
4
1
108
0,9
1,9
0,9
3,7
0,9
100
0
2
0
16
0
191
0
1,0
0
8,4
0
99,9
The subsequent table 4.6 describes the cause of the custom faults. The table also shows the
number of faults at each cause category and how many percent that represents of the total
custom faults.
Table 4.6 Service underground cables
Cause of fault on
custom cables
Lack of maintenance or
worn out cables
Unknown
Unselective
Digging
Incorrect connection
Corrosion
Sabotage
Overload
Fabrication fault
Own fuse
Circuit breaker
Earth fault breaker
Montage fault
Region grid
Traffic
Trees (e.g. roots)
Growth e.g. trees
Lightning
Testing
Wind
Custom fault
Custom ready, by
telephone
Unselective region
Total
Year
2005
(pcs.)
284
Percent of total
custom cable faults
2005 (%)
39,8
Year
2006
(pcs.)
255
Percent of total
custom cable faults
2006 (%)
40,0
169
26
55
2
9
5
43
4
40
5
12
4
2
8
0
0
1
1
1
26
15
23,7
3,6
7,7
0,3
1,3
0,7
6,0
0,6
5,6
0,7
1,7
0,6
0,3
1,1
0
0
0,1
0,1
0,1
3,6
2,1
193
17
52
3
2
9
84
4
0
0
0
8
0
5
1
1
3
1
0
0
0
30,3
2,7
8,2
0,5
0,3
1,4
13,2
0,6
0
0
0
1,3
0
0,8
0,2
0,2
0,5
0,2
0
0
0
2
713
0,3
100
0
638
0
100,4
Table 4.6 shows that in category own fuse there are 40 failures reported in 2005 but zero
failures are reported in 2006. The difference might be that the own fuse failures in 2006 are
reported under the category overload, which almost has twice as much reported failures in
2006 than it had in 2005. The category earth fault breaker and custom faults might also be
reported under other categories in 2006. If this is the case, this shows that the topic of the fault
category causes should be standardised to make it easier to evaluate and compare data from
different years.
62
The tables 4.4-4.6 show that the largest fault category on underground cables is lack of
maintenance or worn out cables. The most possible thing is that the cables is either worn out
or damaged in some way than there is a lack of maintenance, due to the fact that there are
very few cables in the grid that are supposed to be maintained like high voltage cables
insulated by oil. The second large category of caused failures is unknown, which is not very
satisfying. The reason for that could either be that they really are unknown or that there is a
lack of categories in the report data base. The causes of the underground cable failures were
presented in tables 4.4-4.6 for each voltage level. In tables 4.7 and 4.8 the damage to the
system is presented, caused by the different cable faults. This is only made for the high
voltage level, which is of most interest in this report.
Table 4.7 Damage on the power system caused by different causes of failure on high voltage level in 2005
Causes of failures on high voltage level
Pcs. Damage on power system 2005
Lack of maintenance or worn out cables → 45
XLPE-cable
2
Disconnector
1
Connection
7
PILC-Cable
3
System not damaged
2
Transformer
3
Low voltage cable
Total
63
Fabrication fault →
1
Terminations XLPE-cable
Total
1
Digging →
5
XLPE-cable
1
PILC-cable
Total
6
Lightning →
1
System not damaged
Total
1
Unknown →
17
XLPE-cable
3
System not damaged
2
PILC-cable
2
Transformer
1
Connection
Total
25
Unselective-region →
1
XLPE-cable
Total
1
%
71,4
3,2
1,6
11,1
4,8
3,2
4,8
100,1
100
100
83,3
16,7
100
100
100
68,0
12,0
8,0
8,0
4,0
100
100
100
Total high voltage faults in 2005 97
100
Table 4.7 above shows that at the high voltage level the XLPE-cable was the most damaged
component. The XLPE-cables were involved in 69 of 97 high voltage faults in 2005, which is
71 percent of the total high voltage faults. The second most damaged component at the high
voltage level was the PILC-cable. The PILC-cable stood for 10 faults, which are about 10
percent of total high voltage faults in 2005. Totally high voltage cables were the most damage
component in the high voltage system with over 80 percent of the faults in 2005. Of these 80
percent almost 57 percent is caused by lack of maintenance or worn out cables category.
63
Table 4.8 Damage on the power system caused by different causes of failure on high voltage level in 2006
Causes of failures on high voltage level
Pcs.
Lack of maintenance or worn out cables → 42
2
2
1
1
1
1
Total
50
Incorrect connection →
1
1
Total
2
Digging →
11
Total
11
Corrosion →
1
Total
1
Unknown →
19
6
2
1
1
1
Total
30
Unselective →
12
Total
12
Unselective-region →
1
Total
1
Sabotage →
1
Total
1
Overload →
1
Total
1
Damage on power system 2006
XLPE-cable
Joint XLPE-cable
Connection
PILC-cable
System not damaged
Transformer
Termination XLPE-cable
XLPE-cable
PILC-cable
XLPE-cable
XLPE-cable
XLPE-cable
PILC-cable
System not damaged
Connection
Joint XLPE
Transformer
System not damaged
XLPE-cable
XLPE-cable
System not damaged
Total high voltage faults in 2006 109
%
84,0
4,0
4,0
2,0
2,0
2,0
2,0
100
50
50
100
100
100
100
100
63,3
20,0
6,7
3,3
3,3
3,3
99,9
100
100
100
100
100
100
100
100
100
Table 4.8 above shows that at the high voltage level the XLPE-cable was the most damaged
component. The XLPE-cables were involved in 76 of 109 high voltage faults in 2006, which
is almost 70 percent of the total high voltage faults. The second most damage component at
the high voltage level was the PILC-cable. The PILC-cables stood for 8 faults, which are
about 7 percent of total high voltage faults in 2006. Totally high voltage cables were the most
damaged component in the high voltage system with 77 percent of the faults in 2006. Of these
77 percent 50 percent is caused by lack of maintenance or worn out cables category. Most of
the high voltage failures are caused by the cables which are shown in tables 4.7 and 4.8. The
numbers are fairly equal in 2005 and 2006.
4.4.1 Fault frequencies for cables at the high voltage level
The Stockholm grid had 2900 km high voltage cable installed in 2005 and 2006 divided into
30% PILC-cables and 70% XLPE-cables [32]. Table 4.9 shows the fault frequency in form of
faults per hundred-kilometre cable and year for the Stockholm medium voltage grid. The fault
frequency is calculated as formula 4.5 shows.
64
Fault frequency =
Number of faults × 100
Length of cable in kilometre
(4.5)
Table 4.9 Fault frequency in Stockholm’s medium voltage grid in 2005 and in 2006
Fortum’s 12kV and 24kV Stockholm grid
2005 2006
Total length of cables in service (km)
Total faults (pcs.)
The total fault frequency, faults/100 km cable and year
Faults caused by lack of maintenance or worn out cables (pcs.)
Fault frequency, faults/100 km cable and year
Fault caused by digging and other damages (pcs.)
Fault frequency, faults/100 km cable and year
Total length of PILC-cables in service (km)
Total faults directly connected to PILC-cables (pcs.)
Fault frequency, faults/100 km cable and year
Total length of XLPE-cables in service (km)
Total faults directly connected to XLPE-cables (pcs.)
Fault frequency, faults/100 km cable and year
2900
97
3,34
63
2,17
6
0,21
870
10
1,15
2030
68
3,35
2900
109
3,76
50
1,72
11
0,38
870
8
0,92
2030
76
3,74
Table 4.9 above shows that the total fault frequency in faults per100 km cable and year was
3,34 in 2005 and had increased to 3,76 in 2006. The fault frequency is also calculated for the
two largest fault categories, fault caused by worn out cables or lack of maintenance and
digging. The fault frequency in faults per 100 km cable and year for the category lack of
maintenance or worn out cables was 2,17 in 2005 and had decreased to 1,72 in 2006. The
fault frequency in faults per 100 km cable and year for the category digging was 0,21 in 2005
and 0,38 in 2006. The total fault frequency in faults per 100 km cable and year directly
connected to old PILC-cables was 1,15 in 2005 and 0,92 in 2006. The total fault frequency in
fault per 100 km cable and year directly connected to XLPE-cables was 3,35 in 2005 and 3,74
in 2006. However, two years is a too short period of time to state if this is a trend or just a
natural statistic variation.
The data showed in table 4.10 comprises all national failures reported to Darwin during the
period from 2003 to 2005 [33]. Table 4.10 shows the total fault frequency in faults per 100
km cable and year at the 12 kV and 24 kV distribution grids. The fault frequency is calculated
according to formula 4.5.
Table 4.10 Fault frequency reported to Darwin during the period from 2003 to 2005.
The total national 24 kV and 12 kV grid (Darwin)
Total length of cables in service (km)
Total faults (pcs.)
The total fault frequency, faults/100 km cable and year
2003
37156
955
2.57
2004
38666
1115
2.88
2005
58558
1173
2.00
Table 4.10 shows that the average national fault frequency at the 24 kV and 12 kV grids is
considerably lower than Fortum’s medium voltage grid in Stockholm. However, the reliability
is dependent on that all faults are reported to Darwin. A comparison has been made with
Göteborg Energi’s 10 kV grid which is shown in table 4.11.
65
Table 4.11 Fault frequency in Göteborg Energi’s 10 kV grid during the period 1999, 2005-2007
Göteborg Energi’s 10 kV grid
Total length of cables in service (km)
Faults connected to digging (pcs.)
Faults connected to other damages (pcs.)
Faults connected to PILC-cables (pcs.)
Faults connected to XLPE-cables (pcs.)
Faults connected to terminations PILC-cables (pcs.)
Faults connected to terminations XLPE-cables (pcs.)
Faults connected to joint PILC-cables (pcs.)
Faults connected to joint XLPE-cables (pcs.)
Faults connected to joint between PILC- and XLPE-cables (pcs.)
Faults connected to unknown (pcs.)
Total number of faults (pcs.)
The total fault frequency, faults/100 km cable and year
Total length of XLPE-cables in service (km)
Total faults connected to XLPe-cables (pcs.)
Fault frequency, faults/100 km cable and year
Total length of PILC-cables in service (km)
Total faults connected to PILC-cables (pcs.)
Fault frequency, faults/100 km cable and year
1999
1790
10
4
13
5
0
1
20
0
5
2
60
3,35
381
5
1,31
1409
13
0,92
2005
1861
8
2
22
4
0
1
18
0
6
2
63
3,39
490
4
0,82
1371
22
1,60
2006
1892
10
3
22
4
1
1
14
0
9
3
67
3,54
530
4
0,75
1362
22
1,62
2007
1934
10
6
22
7
0
4
13
4
12
0
78
4,03
575
7
1,22
1358
22
1,62
Table 4.11 shows the fault frequency in Göteborg Energi’s 10 kV grid during the period 1999,
2005-2007. Table 4.9 and 4.11 shows that the total fault frequency at Fortum’s Stockholm
medium voltage grid and Göteborg Energi’s medium voltage grid are in the same size.
However, the fault frequency of Göteborg’s PEX-cables is considerable lower than for
Stockholm’s PEX-cables. Although the fault frequency of Göteborg’s PILC-cables are higher
than for Stockholm’s PILC-cables.
4.4.2 Consequences of a failure
Table 4.12 shows some consequences of medium voltage cable failure on the Stockholm grid.
Failure length is the total number of hours that failures have occurred during respective years.
Customer time is the total number of hours that customers have been effected by faults
respectively years. Energy not served is the total amount of energy that Fortum not has
distributed due to cable faults respectively year. Number of effected customers due to cable
faults respectively year.
Table 4.12 Consequences of high voltage failures
Consequence of high voltage failures
Failure length (h)
Customer time (h)
Energy not served (MWh)
Number of effected customers (pcs.)
2005
7165
151 000
305
150 768
2006
547
84 500
316,5
83 573
66
4.5 Statistics from the Ekerö grid 2004 to 2007
Failure statistics were available for Ekerö grid during the period from 2004 to 2007 [34].
There have been 17 medium voltage faults on the Ekerö grid during this period, connected to
underground medium voltage cables. The statistics from the Ekerö grid includes statistics
from Darwin, Ekerö Energi’s failure reports and also some extra information provided by
Henrik Svensson at Ekerö Energi. The failure reports have been examined and compared to
Darwin, in order to get a more detailed view of what have happened. For some of the failures
there was some extra information available that was not reported in the failure reports. The
given information is shown in table 4.13 to table 4.16.
Table 4.13 Underground cable faults in 2004
Failure
report
reg.
number,
date
Station
Year of
cable
installation
Cause of
failure
according
to Darwin
report
Cause of failure
according to
failure report
12,
11 January
2004
17,
17 January
2004
Karlslund
1986
Fault in cable joint
oil/plastic.
Sånga
1993
61,
31 Mars
2004
Lovö
Unknown
Earth-fault.
Fabrication or
material fault
Earth-fault.
Incorrect
installation or
placement.
Earth-fault.
Fabrication or
material fault.
89,
31 May
2004
Brygga
Not interesting
due to the
cause of
failure.
Earth-fault.
Cause of
failure
unknown.
142,
27 July
2004
MSSånga
155,
5 August
2004
Sånga
Not interesting
due to the
cause of
failure.
2000
Over current.
Cause of
failure
unknown.
Earth fault.
Fabrication or
material fault.
67
Extra
information
from Henrik
Svenson’s
(Ekerö Energi)
own notes
Fault in cable
termination.
Short-circuit phase
L3, treeing showed
white spot on the
insulation.
Measurement
showed that the
cable was in bad
condition.
There had been a
flood in the station
and water had rise
up on the
terminations.
Trigged breaker due
to over current.
Cause of failure
unknown.
Cable fault.
Cable fault
indicated treeing.
Table 4.14 Underground cable faults in 2005
Failure
Station
report reg.
number,
date
Year of cable
installation
Cause of
failure
according to
Darwin
report
Cause of
failure
according to
failure
report
103,
25 April
2005
NS1080
Digging
203,
15
November
2005
Munsö
Not interesting
due to the
cause of
failure.
1954
Digging. The
cable is not
properly
marked.
Cable fault on
a sea
reinforced seacable.
228,
5 December
2005
Träkvista 1965
Earth fault on
sea-cable.
Incorrect
installation or
placement.
Over current.
Fabrication or
material fault.
Cable fault.
Extra
information from
Henrik
Svenson’s (Ekerö
Energi) own
notes
Sea reinforced seacable. Cable fault
ice and gravel had
damaged the cable
Oil leakage on
PILC-cable at joint
between PILC and
XLPE.
Table 4.15 Underground cable faults in 2006
Failure
report
reg.
number,
date
Station
34,
5 February
2006
Jehander
53,
23 February
2006
96,
22 May
2006
Year of
cable
installation
Not interesting
due to the
cause of
failure.
Träkvista Not interesting
due to the
cause of
failure.
Sånga
Not interesting
due to the
cause of
failure.
107,
20 June
2006
Lovö
Unknown
176,
23 October
2006
Sånga
Unknown
189,
31 October
2006
Sånga
1974
Cause of
failure
according to
Darwin
report
Cause of failure
according to
failure report
Customer
fault.
Not interesting
due to the failure
cause.
Over current.
Digging
Digging.
Not
properly marked.
Extra
information
from Henrik
Svenson’s
(Ekerö Energi)
own notes
Over current. No fault. The
Unknown.
breaker
had
probably trigged
when another able
was reconnected
on another line.
Earth
fault. Earth fault.
Fabrication
or
Fabrication or
material
fault.
material fault.
Cable treeing filled
with water.
Over current. Cause unknown.
Unknown.
Breaker trigged on
all three phases for
over current.
Over current. Cable
fault
Fabrication or damaged
material fault. termination.
68
227,
15
December
2006
Not interesting Over current. Digging.
due to the Digging
cause
of
failure.
Lovö
Table 4.16 Underground cable faults in 2007
Failure
Station
report reg.
number,
date
Year of
cable
installation
Cause of
failure
according to
Darwin
report
Cause of
failure
according to
failure report
Extra information
from Henrik
Svenson’s (Ekerö
Energi) own notes
77,
30 April
2007
1983
Earth fault.
Fabrication or
material fault.
Earth fault.
Fault in joint
between PILC and
XLPE.
MSBrygga
Table 4.17 Summary of failures reported to Darwin and Ekerö Energi’s own failure reports from 2004 to 2007.
Cause of failure
according to
Darwin
Pieces Cause of failure
according to failure
report
Pieces Extra information from
Henrik Svenson’s (Ekerö
Energi) own notes
Fabrication or
material fault
7
Cable joint fault
1
Treeing
Cable fault
1
2
Earth fault
2
Termination fault
There had been a flood in the
station and water had rise up
on the terminations.
Unknown
Digging. Not properly
marked
Digging
Fault in cable termination
1
1
Cable fault on a sea
reinforced sea-cable.
1
Customer fault
1
17
Unknown
4
Digging
3
Incorrect
installation or
placement
2
Customer fault
Total
1
17
Cable fault indicated treeing
Oil leakage on PILC-cable at
joint between PILC and
XLPE.
Fabrication or material fault.
Cable treeing filled with
water.
Fault in joint between PILC
and XLPE.
3
1
2
1
Sea reinforced sea-cable.
Cable fault ice and gravel had
damaged the cable
Table 4.17 summarises the failure reports at Ekerö Energi and what was reported to Darwin
during the period from 2004 to 2007. During this period seven faults were reported as
69
material or fabrication faults to Darwin. When the information was examined one step down
in the chain at Ekerö Energi’s failure reports. It showed that these seven faults, which were
reported under one category in Darwin, could be divided into five different fault categories.
Even more fault information was available at internal notes at Ekerö Energi. This notes
showed that two of the fault categories reported in Ekerö Energi’s failure reports could be
divided into four different faults.
The table 4.18 shows the fault frequency in Ekerö’s medium voltage grid during the period
from 2004 to 2007. The fault frequency calculations below are made with formula 4.6.
Fault frequency =
Number of faults × 100
Length of cable in kilometre × Number of years
(4.6)
Table 4.18 Fault frequency in Ekerö’s medium voltage grid from 2004 to 2007
Ekerö’s medium voltage grid from 2004 to 2007
Total length of cables in service (km)
Total faults (pcs.)
The total fault frequency, faults/100 km cable and year
Faults caused by digging and other damages (pcs.)
Fault frequency, faults/100 km cable and year
Total length of XLPE-cables in service (km)
Faults caused by insulation failures on XLPE-cables (pcs.)
Fault frequency, faults/100 km cable and year
2004-2007
214
17
1,99
4
0,47
187
3
0,4
Table 4.18 shows that the total fault frequency in faults per 100 km cable and year was 1,99
during the period from 2004 to 2007. Fault frequency caused by digging and other damage,
which was the largest fault category, was 0,47 faults per 100 km cable and year during this
period. The total fault frequency on XLPE-cables connected to insulation faults was 0,4 faults
per 100 km cable and year during this period.
Finally a comparison of the total fault frequencies between Stockholm, Göteborg, Ekerö and
the faults reported to Darwin has been made. The fault frequency has been calculated for
available years at each grid, except for year 1999 at the Göteborg grid. The comparison is
shown in table 4.19 and the calculations have been made with formula 4.6.
70
Table 4.19 The total fault frequency comparison between Stockholm, Göteborg, Ekerö and Darwin
Fortum’s 12kV and 24kV Stockholm grid
Total length of cables in service each year (km)
Total faults (pcs.)
The total fault frequency, faults/100 km cable and year
Göteborg Energi’s 10 kV grid
Total cable length of cables in service each year (km)
Total faults (pcs.)
Fault frequency, faults/100 km cable and year
Ekerö’s medium voltage grid from 2004 to 2007
Total length of cables in service each year (km)
Total faults (pcs.)
Fault frequency, faults/100 km cable and year
The total national 24 kV and 12 kV grid (Darwin)
Total average length of cables in service each year (km)
Total faults (pcs.)
Fault frequency, faults/100 km cable and year
2005-2006
2900, 2900
206
3,55
2005-2007
1861, 1892, 1934
268
3,58
2004-2007
214, 214, 214, 214
17
1,99
2003-2005
37156, 38666, 58558
3243
2.41
Table 4.19 shows that the total fault frequency at both Stockholm and Göteborg are in the
same size and considerable higher than the total fault frequency calculated from the data
reported to Darwin. Table 4.19 also shows that Ekerö has even lower total fault frequency
than Darwin.
71
5 Discussion
The measurements of the cables at the Ekerö grid show that some of them are aged but not
critical. Most of the cables are in rather good condition, even though some of them are almost
40 years. Of the measured cables are the cables from the 70s in worst condition. However,
they will probably still be in service at least for 5-10 more years. Ekerö Energi had only had
three cable breakdowns related to cable insulation the last four years. At least one of the
cables was placed in clay soil and had damage on the jacket. Besides there are very good
conditions to place cables in the soil at Ekerö, due to the fact that the soil is consisting of sand
and gravel, which has a draining quality. According to Mikael Olofsson, electrician at Ekerö
Energi since 20 years, another reason that Ekerö Energi has avoided breakdowns by water tree
deteriorated cables could have been that they have used a lot of sand when placing the cables
in the ground.
The measured cables could be divided into four groups. In the first group there are cables
from the early 70s, cable #2, #5 and #6. In the second group there are cables from the early
80s, cable #1 and #4. In the third group there is a cable from the mid 80s, cable #7. In the last
group there is a cable with two joints where one part is from the early 70s and the other two
parts are from the mid 80s, cable #3.
The first group of cables from the early 70s, cables #2, #5 and #6 has varying analysis. Cable
#2 was difficult to comment on at all due to creep currents. The reason for creep currents is
most likely that the cable was not disconnected from the isolators. However, the capacitance
plots show no voltage dependence, which indicates that the cable is in good condition. The
cable will remain in service for many years. Cable #5 is aged and probably deteriorated by
water trees. Phase L1 is more deteriorated than phases L2 and L3 that have lower losses. The
cable will probably remain in service in many more years. Cable #6 showed VDP response
but no hysteresis in the tangents delta plot. The capacitance plots showed almost no voltage
dependence. Small water trees probably deteriorate the cable. The cable will remain in service
for many more years.
The second group of cables is from the early 80s, cables #1 and #4. Cable #1 does not show
any voltage dependence and no hysteresis, the losses are also low. The capacitance plots do
not show any voltage dependence. This cable is in good condition and will remain in service
for many years. Cable #4 is difficult to analyse, due to the fact that the cable was not
disconnected from the isolators. This probably created a leakage current response. The
capacitance plots show no voltage dependence, which indicates that this cable is in good
condition.
The third group is a cable from the mid 80s, cable #7. This cable does not show any voltage
dependence and hysteresis response. The losses are rather high probably due to a joint. The
capacitance plots show no voltage dependence. The cable is most likely in a good condition
and not deteriorated by water trees. The cable will remain in service for many years.
The last group is a cable with two joints, cable #3. At both ends of the cable there are cables
from the mid 80s and in the middle there is a cable from the beginning of the 70s. This cable
shows a voltage independent leakage current response with no hysteresis. The capacitance did
not show any voltage dependence either. This indicates that the cable is in good condition and
not deteriorated by water trees. The leakage current response is most likely an effect of the
two cable joints. The cable will probably remain in service for many years.
72
Regarding the measurements made in 1998 it is difficult to say how reliable they were. What
could be said on the other hand, is that even though three of the cables were judged as old and
four of the cables were judged as critical for nine years ago they are still operating.
The insulation diagnostic instruments that were found in the literature study are all off-line
instruments. The technique used in the instruments varies. General Electric’s IDA 200 HVU,
now manufactured and sold by Pax diagnostics and called Idax-206 VAX-230, measures the
dissipation factor tan δ at a variable frequency sweep between 0.0001-100 Hz. Bauer’s PHG
70 and PHG 80 and sebaKMT’s VLF and HV diagnostics HVA 30, 30,5 and 60 measures the
dissipation factor tan δ at a fixed very low frequency, VLF. Tettex Midas, 2820 and 2877
measures the dissipation factor tan δ at power-line frequency. Omicron’s CPC 100 with CP
TD1 also measures the dissipation factor tan δ , however, at a frequency sweep between 15-70
Hz. sebaKMT’s CDS measures the losses with the isothermal relaxation current, IRC method.
The measurement techniques used in these instruments are considered to be non-destructive at
a voltage level up to U 0 .
The fault location instruments found in the literature study are all based on the same
technology, the time domain reflectometry. SebaKMT provides four instruments the
Surgeflex 15, 25, 32 and the Teleflex T-30E. Surgeflex 15 and 25 and Teleflex T-30E are
battery-powered and portable instruments. PFL40-1500/2000 is supplied by Megger and is a
battery-powered portable instrument. The 5150 first response cable fault locator by
Hipotronics is also portable and battery-powered.
The failure statistics from the Stockholm grid shows that the major number of faults is
connected to underground cables. The largest fault category for high voltage, low voltage and
service cables are lack of maintenance or worn out cables. The second large category for all
three-voltage levels is unknown. The largest damages on the high voltage system caused by
failures in these two categories are connected to the XLPE-cables. However, reading the
statistic it is not possible to say if the XLPE-cable failures are connected to insulation
problems or something else. The fault frequency in Stockholm’s XLPE-cables was 3,34 in
2005 and 3,76 in 2006, which is considerable higher than Göteborg Energi’s fault frequency
for XLPE-cables. The fault frequency in Stockholm’s PILC-cables was 1,15 in 2005 and 0,92
in 2006, which is slightly lower than Göteborg Energi’s fault frequency for PILC-cables.
A more extensive investigation of the failure statistics from the Ekerö grid was made. The
statistics reported to Darwin, Ekerö Energi’s own statistics and the failure reports that the
statistics are based on was investigated. Extra information in some faults was also available.
The fault frequency in Ekerö’s XLPE-cable grid connected to insulation failures was 0.4
faults per 100 km cable and year, which is considered to be low. The fault frequency of total
number of faults was 1,99, which is considered to be a low failure frequency compared to
Stockholm and Göteborg. Ekerö’s total fault frequency is also lower than the total fault
frequency reported to Darwin, which is 2,41. However, Ekerö’s fault frequency is a bit
misleading due to the fact that digging, customer fault and a flood in the station cause five of
17 faults. If the fault frequency is recalculated only with faults related to the cable or cable
accessories the fault frequency per 100 km cable and year would be 1.4.
Darwin is a national database to which the energy companies report their failure statistics.
Table 4.17 summarises the failure reports at Ekerö Energi and what was reported to Darwin
during the period from 2004 to 2007. During this period seven faults were reported as
material or fabrication faults to Darwin. When the information was examined one step down
73
in the chain at Ekerö Energi’s failure reports. It showed that these seven faults, which were
reported under one category in Darwin, could be divided into five different fault categories.
Even more fault information was available at internal notes at Ekerö Energi. This notes
showed that two of the fault categories reported in Ekerö Energi’s failure reports could be
divided into four different faults. These examples show that a lot of information is lost during
the report procedure. To be able to get reliable information and to get an efficient use of
Darwin there has to be more categories inserted. Today it is impossible to separate for
example treeing from a joint fault or a termination fault due to the fact that they are reported
under the same category.
Today the failures are reported on paper sheets and saved in paper form at Ekerö Energi. The
quality of the report is very much dependent on the writer. To write down not just the effect
of a fault but also the cause gives valuable information for future investigations. It would be
easier to study failures in a later period of time if the failure reports were written directly into
a database. Another thing that would increase the knowledge about the condition of the
medium/high voltage grid, would be if there was established a protocol or a journal of each
medium/high voltage cable. This protocol or journal should give essential information about
the cable. Such as manufacturer, type and which year the cable was manufactured. The
protocol or journal should give information about type and manufacturer of the terminations
and any joints. There should also be information about failures cause and effect. If there had
been any diagnostic measurements there should be information about these.
74
6 Conclusion
The aim with this report was to:
•
•
•
•
•
perform cable diagnostic on Ekerö Energi’s medium voltage grid.
analyse the diagnosed cables at Ekerö.
try to evaluate if the measurements made in 1998 were trustworthy.
give an overview of different commercial XLPE-cable diagnostic instruments
and fault location instruments.
analyse the failure statistic at the Stockholm grid.
The subsequent paragraphs give the results.
The measurements of the cables at the Ekerö grid show that some of them are aged but not
critical. Most of the cables are in rather good condition and are suggested to be re-measured
in 5 to 10 years. However, the cables will probably remain in service for many years.
Regarding the measurements made in 1998 it is difficult to say how reliable they were.
However, they are still in service.
The insulation diagnostic instruments for diagnosis of XLPE-cables available on the market in
the beginning of 2008 are all off-line instruments. All of them except for one measure the
dissipation factor tanδ at either a variable low frequency or at a fixed low or grid frequency.
The other method used is to measure the Isothermal Relaxation Current IRC. All instruments
are considered to use non-destructive methods for voltage levels up to U 0 . All fault location
instruments revealed in this report use Time Domain Reflectometry TDR.
The failure statistics in 2005 and in 2006 from the Stockholm grid shows that the major
number of faults is connected to underground cables. However, the period of two years is too
short to draw any conclusions on the variations between the years. The fault frequency in
Stockholm’s XLPE-cables was 3,34 in 2005 and 3,76 in 2006, which is considerable higher
than Göteborg Energi’s fault frequency for XLPE-cables. The fault frequency in Stockholm’s
PILC-cables was 1,15 in 2005 and 0,92 in 2006, which is slightly lower than Göteborg
Energi’s fault frequency for PILC-cables. However, the total fault frequency is in the same
size for Stockholm and Göteborg.
The failure statistics from Ekerö’s XLPE-cable grid shows that the fault frequency of total
number of faults was 1,99. This is considered to be a low failure frequency compared to
Stockholm and the failure frequency reported to Darwin. However, the fault frequency
connected to insulation failures at Ekerö was 0.4 faults per 100 km cable and year, which is
considered to be low.
Darwin is a national database to which the energy companies report their failure statistics. To
be able to get reliable information and to get an efficient use of Darwin there has to be more
categories inserted. Today it is impossible to separate for example treeing from a joint fault or
a termination fault due to the fact that they are reported under the same category.
75
Appendices
Appendix A
List of symbols and abbreviations
Symbol
A
ARM
C
CW
C0
C
C′
C ′′
D
d
E
ε = εr ε 0
ε r′
ε r′′
ε0
f (t )
G
I
Ic
Id
Il
ICM
J
R
R0
RW
σ
TDR
τ
tan δ
U=V
U0
ω
Symbol explanation
Surface area
Arc Reflection Method
Capacitance
Capacitance over water tree
Capacitance in vacuum
Complex capacitance
True capacitance
Represents the lossy part
Displacement factor
Thickness of the dielectric
Electric field
Complex permittivity
Real part of the permittivity
Imaginary part of the relative complex permittivity = dielectric loss
factor
Permittivity value in vacuum
Response function
Conductance
Current
Charging current
Depolarisation current
Leakage current
Impulse Current Method
Current density
Resistance
DC-resistance
Water tree resistance
Conductivity
Time Domain Reflectometry
Time constant
Dissipation factor (Loss tangent)
Voltage
System phase to earth voltage
Angular frequency
76
Acronyms and abbreviations
AC
Alternating current
CDS
Cable Diagnostic System
DC
Direct current
DSP
Digital Signal Processing
HV
High Voltage
HVU
High Voltage Unit
IRC
Isothermal Relaxation Current
KTH
Kungliga Tekniska Högskolan (Royal Institute of Technology),
Stockholm
L
Line (Phase 1 = L1)
LC
Leakage current
MS
Receiver Station (Mottagar station)
NTNU
Norges Tekniska och Naturvetenskapliga Universitet
NS
Grid Station (Nät Station)
PD
Partial Discharge
PDC
Polarisation / Depolarisation Current
PE
Polyethylene
PHG
Programmable High voltage Generator
PILC
Paper Insulated Lead Covered cable
RMS
Root Mean Square
RH
Relative Humidity
RVM
Return Voltage Method
SEfAS
SINTEF Energis forsknings Aktie Sellskap
TD
Time Domain or dissipation factor measurement
TDR
Time Domain Reflectometry
TLC
Transition to Leakage Current
VDP
Voltage Dependent Permittivity
VLF
Very Low Frequency
XLPE
Cross-Linked Polyethylene
77
Appendix B
Matlab code for the Bush tree models
clc
clear all
close all
fstart=0.0001
fslut=100
h=0.0001
Rwbush=2e14;
Rwleakage=2.5e10;
Co=2.5e-10;
Cw=3e-13;
Num_of_trees_per_meter=100;
num_of_points=100;
f=logspace(log10(fstart),log10(fslut),num_of_points);
w=2*pi*f;
tau=Rwbush*Cw;
Cbisbush=Num_of_trees_per_meter*w*Rwbush*Cw^2./(1+(w*tau).^2);
Cprimbush=Co+Num_of_trees_per_meter*Cw./(1+(w*tau).^2);
Cbisleakage=1./(Num_of_trees_per_meter*w*Rwleakage);
Cprimleakage=Co;
tandbush=Cbisbush./Cprimbush;
tandleakage=Cbisleakage./Cprimleakage;
figure(1)
loglog(f,tandleakage),xlabel('Frequency Hz'),ylabel('tandelta')
title('Model of leakage tree deteriorated cable')
figure(2)
loglog(f,tandbush)
ylabel('tandelta')
xlabel('Frequency Hz')
title('Model of bush tree deteriorated cable')
78
Appendix C
Pictures from measurements made on the Ekerö grid in 2007
Figure C.1. Measurement performed with IDA 200 HVU at Klyvarestigen. The measurement
was performed at cable #3 between NS 2123 Klyvarestigen and NS 2126 Uppgårdsskolan.
Figure C.2. Measurement performed with on-voltage TDR at Klyvarestigen. The measurement
was performed at cable #3 on phase 1 between NS 2123 Klyvarestigen and NS 2126
Uppgårdsskolan.
79
References
[1] SCB Statistiska Central Byrån, Tillförsel och användning av el 1996-2006 (GWh),
Documenting Electronic Sources on the Internet. 01 Oct. 2007
<http://www.scb.se/templates/tableOrChart____24270.asp> (Accessed 08 Jan. 2008)
[2] P. Werelius, Development and Application of High Voltage Dielectric Spectroscopy for
Diagnosis of Medium Voltage XLPE Cables. Ph.D. Thesis, Kungliga Tekniska Högskolan,
Department of Electrical Engineering, Stockholm, 2001 TRITA-ETS-2001-02, ISSN 1650674x
[3] R. Bartnikas, K.D. Srivastava (2000), Power and Communication Cables, (p. 332-335),
Institute of Electrical and Electronics Engineers, Inc. 3 Park Avenue, 17th Floor, New York,
NY 10016-5997, IEEE ISBN 0-7803-1196-5, IEEE Order Number PC5665
[4] S. Hvidsten, Tilstandskontroll av fire kommersielle metoder, ISBN 82-594-1673-5,
SINTEF Energiforskning AS, Sem Saelands vei 11, 7465 Trondheim, Norway, November
1999
[5] T. K. Saha, P. Purkait, Investigating some important parametres of the PDC measurement
technique for the insulation condition assessment of power transformer, University of
Queensland Australia, (2003).
[6] T. K. Saha, P. Purkait, Condition monitoring of transformer insulation by polarization and
depolarization current measurements, School of Information Technology and Electrical
engineering, University of Queensland, Brisbane, QLD-4072, Australia
[7] V. Dubickas, H. Edin, R. Papazyan, Cable Diagnostics with On-voltage Time Domain
Reflectometry, Stockholm, Royal Institute of Technology, Department of Electrical
Engineering, NORDAC-2006
[8] GE Energy Services, Programma products, IDA 200 HVU Insulation Diagnostic System,
Documenting Electronic Sources on the Internet.
<http://www.interfaxsystems.com/interfaxusa/Datasheets/Testing%20Injection%20Testing/id
a200_en.pdf> (Accessed 26 Sep. 2007)
[9] Pax Diagnostics, Idax-206 Insulation Diagnostics Analyze, Documenting Electronic
Sources on the Internet.
<http://www.paxdiagnostics.net/uploads/Idax206.pdf> (Accessed 11 Dec. 2007)
[10] Pax Diagnostics, VAX-230 High Voltage Amplifier, Documenting Electronic Sources on
the Internet. 19 Oct. 2007
<http://www.paxdiagnostics.com/uploads/VAX-230_A4_071019.pdf> (Accessed 11 Dec.
2007)
[11] B. Oyegoke, P. Hyvönen, M. Aro (2001), Dielectric Response Measurements as
Diagnostic Tool for Power Cable Systems, Espoo, Finland, Helsinki University of
Technology, High Voltage Institution Report TKK-SJT-47, ISSN 1237-895x, ISBN 951-225396-8
80
[12] Baur, Documenting Electronic Sources on the Internet. 29 May 2007.
<http://www2.baur.at/fileadmin/pdf/HSP/en_826-066_PHG7080_TDPD.pdf> (Accessed 27
Sep. 2007)
[13 sbaKMT, CDS-Cable Diagnostic System for three phase IRC and RVM diagnostics,
Documenting Electronic Sources on the Internet. 2007.
< http://www.sebakmt.com/fileadmin/dam/LFT_CDS_eng_2007_08.pdf > (Accessed 28 Sep.
2007)
[14] H.-G. Kranz, D. Steinbrink, Bergische Universität-GH Wuppertal Germany, F.Merschel,
RWE Energie AG Essen, Germany, IRC-analysis: Anew test procedure for laid XLPE-cables,
10th International Symposium on High Voltage Engineering, Montréal, Québec, Canada, 2527 August, 1997
[15] M. Beigert, H.-G. Kranz, Bergische Unversität-GH Wuppertal, Germany, D. Kaubisch,
D. Meurer, KABEL HEYDT AG, Diusburg, Germany, Computer-Aided Destruction Free
Ageing Diagnosis For Medium-Voltage cabels, 8th International Symposium on High Voltage
Engineering, Yokohama, Japan, 23-27 August, 1993
[16] SebaKMT, VLF Test Systems 28-60 kV Portable test systems to generate 0.1 HZ VLF AC
and DC test voltages, Documenting Electronic Sources on the Internet. 2005
<http://www.sebakmt.com/fileadmin/user_upload/pdfs/eng/LFT_VLF28_40_60_eng_2005_3
0.pdf > (Accessed 30 Oct. 2007)
[17] Tettex Instruments, MIDAS Mobile Insulation Diagnosis & Analysing System,
Documenting Electronic Sources on the Internet. Jul. 2007
<http://www.haefely.com/pdf/LL_MIDAS288x_0707RS_l.pdf> (Accessed 01 Oct. 2007)
[18] Tettex Instruments, 2820 Automated C, L & tan δ / cos φ measuring bridge, Documenting
Electronic Sources on the Internet. Jul. 2007
<http://www.haefely.com/pdf/LL_2820_0707RF_l.pdf> (Accessed 01 Oct. 2007)
[19] Tettex Instruments, 2877 Automated C, L & tan δ / cos φ precision measuring bridge,
Documenting Electronic Sources on the Internet. Jul. 2007
<http://www.haefely.com/pdf/LL_2877_0707RF_l.pdf (Accessed 01 Oct. 2007)
[20] HV Diagnostics, HVA 60 4 in 1Universal High Voltage Test System, Documenting
Electronic Sources on the Internet. 27 Sep. 2005
<http://www.hvdiagnostics.com/PDF/HVA60%20DataSheet.pdf> (Accessed 01 Oct. 2007)
[21] Omicron, Primary Testing CP-LINE CATALOG, Documenting Electronic Sources on the
Internet.
<http://www.omicron.at/fileadmin/user_upload/files/pdf/en/CPLine-ENU-6.2.1.0-Lr.pdf>
(Accessed 09 Jan. 2008)
[22] SebaKMT, Surgeflex 15, Surgeflex 25 Mobile Battery Operated Cable Fault Locating
System, Documenting Electronic Sources on the Internet. 2005
< http://www.sebakmt.com/fileadmin/dam/LFT_surgeflex15_25_eng_2005_07.pdf >
(Accessed 11 Jan. 2008)
81
[23] SebaKMT, Surgeflex 32 Portable cable test and fault location system up to 32 kV for
medium and low voltage power cables, Documenting Electronic Sources on the Internet. 2007
< http://www.sebakmt.com/fileadmin/dam/LFT_E_surgeflex32_eng_2007_26.PDF >
(Accessed 11 Jan. 2008)
[24] SebaKMT, Teleflex T 30-E Time DomainReflectometre, Documenting Electronic Sources
on the Internet. 2007
< http://www.sebakmt.com/fileadmin/dam/LFT_T30E_eng_2007_08.pdf> (Accessed 11 Jan.
2008)
[25] Megger, PFL40-1500/2000 Portable Cable Fault Location and High Voltage Test
Solutions, Documenting Electronic Sources on the Internet.
< http://www.megger.com/eu/products/ProductDetails.php?ID=866&Description= >
(Accessed 11 Jan. 2008)
[26] Alibaba, Cable Fault Locator, Documenting Electronic Sources on the Internet. 2008
<http://www.alibaba.com/catalog/10800957/Cable_Fault_Locator.html> (Accessed 09 Jan.
2008)
[27] SebaKMT, Dielectrische Diagnose Geräte KDA-1 und CD-3, Documenting Electronic
Sources on the Internet. 2001
<http://www.sebareseaux.fr/pdf/42_d.pdf> (Accessed 28 Sep. 2007)
[28] University of Washington, College of Engineering, Electrical Engineering. Documenting
Electronic Sources on the Internet.
<http://www.ee.washington.edu/.../sensors/waterTree3.jpg> (Accessed 11 Jan. 2008)
[29] Picture from assistant professor Hans Edin, KTH, Sweden
[30] Johansson Lindfors Maj-Britt, Att utveckla kunskap – Om metodologiska och andra
vägval vid vetenskaplig kunskapsbildning, Studentlitteratur, Lund, 1993
[31] Wallen Göran, Vetenskapsteori och forskningsmetodik, (2:a uppl.), Studentlitteratur,
Lund, 1996
[32] Olle Hansson, Fortum, Stockholm, Sweden. (Private correspondence)
[33] Matz Tapper, Svensk Energi, Stockholm, Sweden (Private correspondence)
[34] Henrik Svensson, Ekerö energi, Ekerö, Sweden. (Private correspondence)
82
Bibliography
L. Bertling (2002), Reliability Centered Maintenance for Electric Power Distribution
Systems, Ph.D. Thesis, Stockholm, Royal Institute of Technology, Department of Electrical
Engineering TRITA-ETS-2002-01, ISSN 1650-674x, ISBN 91-7283-345-9
T. Lindquist, Nord-Is 2003
R. Eriksson, T. Lindquist, L. Bertling, Reliability modeling of aged XLPE cables, Presented at
Nordic Insulation Symposium Tampere, 11-13 June, 2003
R. Eriksson, Requirement on Power Supply Reliability and its Influence Upon the Design of
Electricity Distribution, Stockholm, Royal Institute of Technology, Department of Electrical
Engineering
J. Heggeset1, E. Solvang1, L. Bertling2, J.S. Christiansen3, H. Engen4, K. R. Backen5, J.
Pylvänäinen6, J Hasselström7, Failure Models for Network Components as a Basis for Asset
Management, SINTEF Energy Research1, KTH2, DEFU3, Stattnet4, Tampere University of
Technology5, Hafslund Nett6, Fortum7, NORDAC-2006
E. Kuffel, W. S. Zaengl, J. Kuffel, “High Voltage Engineering: Fundamentals” Second
edition ISBN 0 7506 3634 3, Biddles Ltd, Great Britain, 2000
J. Setréus, l. Bertling, S.M. Gargari, Simulation Method for Reliability Assessment of
Electrical Distribution Systems, Stockholm, Royal University of Technology, Department of
Electrical Engineering
O. Vähämäki1, S. Sauna-aho1, S. Hänninen2, M. Lehtonen3, A New Technique for Short
Circuit Fault Location in Distribution Networks, Finland, Vamp Ltd, Vaasa1, VTT Energy,
Espoo2, Helsinki University of Technology, Espoo3, NORDAC-2006
83
XR-EE-ETK 2008:001
www.kth.se