1

Abstract- Condition assessment of HV assets is one of the issues of asset management introduction in power utility business. In particular, due to there importance in the transmission network is the knowledge about the starting conditions during after-laying as well as the actual condition of HV power cable sections during operation after several years of service of great importance.

In an international co-operation, based on utility experiences and laboratory investigations as obtained for PD diagnosis of distribution power cables using damped AC voltages a complete new method of PD detection and localization for transmission power cables up to 250kV has been developed in 2003 and in use for already one year. In this contribution based on field experiences as obtained in The Netherlands for cable systems up to 150kV the aspects of sensitive PD detection and location are presented and discussed.

Index Terms—HV power cables, on-site, diagnosis, condition assessment, asset management.

I. GENERAL From the utility point of view the reliability and availability of connections of the transmission network are very important. In particular, the classification of cable condition to support the Asset Management decisions abut maintenance policy is the major goal of on-site testing of service aged HV cables, figure 1. It is known, that the insulation failures in a cable network may be caused by lower dielectric strength due to aging processes and by internal defects in the insulation system. I. To reduce the failures by the aging of the impregnated

insulation different types of diagnostic methods are in use: (a) assessment of bulk properties of insulation e.g. dielectric losses measurements (b) return voltage measurement.

II. To reduce the failure by internal defects, on-site cable diagnostics can be applied based on quantities related to insulation degradation, as partial discharges.

After a survey of all relevant information about the cables in the network, diagnostics are carried out to assess the condition of each cable. Interpretation is done based on criteria for each diagnostic. All results together are used for a classification of

1 Delft University of Technology, The Netherlands (e.gulski@ewi.tudelft.nl) 2 Seitz Instruments AG, Switzerland (p.p.seitz@seitz-instruments.ch) 3 SebaKMT, Germany (frank.petzold@sebakmt.com) 4 KSANDR, The Netherlands (e.r.s.groot@ksandr.org)

the cable into four possible categories, see table 1. It follows from this table that based on information which is provided during periodic or condition based inspection the actual condition of a cable section can be used to plane the necessary maintenance activities and to determine the reliability of this particular section in the total network configuration.

II. POWER CABLES DIAGNOSIS TOOLS Unlike voltage testing, measurements of the dielectric properties are valuable indicator for the quality level of the cable insulation. The results of these measurements have a direct relation to the average qualitative level of the insulation at the moment of measurement and can thus be applied as a trend- or fingerprint measurement. Partial discharges are an indication for weak spots in a cable connection. In order to run the measurement partial discharges are ignited in the cable insulation or joints by the application of a test voltage [4]. Due to the physical character of discharge occurrence, such as the PD inception voltage, the PD pulse magnitudes versus voltage applied, PD patterns and PD

On-site Diagnosis of Transmission Power Cables

1Edward Gulski Senior Member, IEEE, 2Paul P. Seitz, 3Frank Petzold, 4Edwin R.S. Groot

Fig. 1. - Examples of condition assessment of 150kV Gas-pressurized mass-insulated power cable, The Netherlands. Using advanced technologies OWTS250, CDS the integral condition and the weak spots location are applied to classify the cable condition.

2

mappings for a utility interested in applying PD diagnostics for condition assessment of its power cable networks, a number of technical and economical aspects are of importance: Voltage type: equivalence in PD inception processes among different voltage stresses for solid insulating materials; Non-destructiveness: non-destructiveness of voltage stress during the diagnosis; IEC 60270 conformity: in the case of measuring the PD quantity apparent charge of PD pulses in [pC] and [nC] the PD detection methods applied has to fulfill the recommendation of IEC 60270; Sensitivity: immunity for on-site interferences and the level of system background noise; Analysis: possibility to generate advanced diagnostic information to support diagnostic knowledge rules; Efficiency: investment costs, maintenance costs, transportability and operation of the method in different field circumstances.

TABLE I TILITY GOALS WITH REGARD TO CLASSFIFCATION OF CABEL

CONDITION ASSET MANAGEMENT PURPOSES. Condition Definition Normal No problems

No maintenance necessary Fault initiation

Short term: No impact on network reliability Long term: without any maintenance possible life time reduction

Fault Short term: cable can still be operated but the network reliability is decreased Maintenance is necessary

Failure Cable can not be operated and maintenance is necessary Based on economics repair or replacement

The dielectric losses measurement is applied for the determination of the loss factor of the insulation material. This factor increases during the ageing process of the cable. This value of a cable is strongly influenced by the composition of the connection, the trace, and the deviations in joints and the actual cable temperature. The dielectric losses measurement is only applicable as trend measurement if composition circumstances of the trace and thermal conditions of successive measurements are virtually identical. The dielectric losses measurement is not applicable for XLPE cables due to the very low value. For HV paper insulated cables and mass-impregnated cables the dielectric losses measurement can be an important indicator of possible thermal breakdowns [2, 3]. Return voltage measurement as used to paper-oil and mass-insulation may provide information about the degradation status of the insulation. All aging processes change the way a dielectric responds to the application of an electric field. Based on the relation between aging and polarization processes the shape of return voltage depends on active polarization mechanisms.

III. TOOLS FOR ON-SITE DIAGNOSIS HV POWER CABELS

A. Partial discharges and dielectric losses To perform diagnostic measurements the HV power cable has

to be energized. Unfortunately, the use of on-line PD analyzing techniques (known from after laying tests of XLPE cable accessories) is limited to detecting at U0 voltage level PD activity in cables accessories only. With regard to the whole section no information can be obtained about the PD behavior at different voltages and the no insight is given into the cable insulation degradation. For the on-site detection of PD related defects in power cables, it is necessary to energize the disconnected cable sample for the ignition of the PD sources. The detection equipment is therefore directly connected to the cable conductors (or through the switchgear). In this way, the different phases of the cable circuit can be energized and the PD pulses can be detected and analyzed. The capacitive power P needed to stress on-site the cable insulation is determined by 2Π•f•Ccable•U2

test where f is the test voltage frequency, Ccable is the cable capacitance and Utest, is the test voltage level. In order to decrease the capacitive power demands for energizing cables as compared to 50 Hz test voltages, a new solution for PD diagnosis of transmission power cables is in use since 2004 in Europe. This solution covers the on-site application of damped AC (DAC) voltages in the range of 50Hz up to 500Hz to energize HV cables and to detect and to locate discharging sites. The off-line PD diagnostic tool as used for the investigations in this paper is based on external energizing of a cable circuit by damped AC voltages with a voltage source up to 250kV. Originally, the DAC voltages are introduced as a cost-effective withstand-voltage test for XLPE-insulated HV cables. Nowadays, DAC voltages are more and more used for non-destructive PD diagnosis of distribution power cables [2, 4, 5]. One of the methods using DAC voltages for detection and localization of PD in cables is known as Oscillating Wave Test System (OWTS), see figures 1 and 2 [4,5]. For the generation of damped AC voltages, the power demand is low due to the charging the cable capacitance with an continuously increasing HV stress, after which the cable capacitance is switched in series with large inductance, resulting in an oscillating voltage wave with a frequency comparable to power frequencies. Figure 2 shows a schema of the energizing method for damped AC voltages for PD detection in power cables. The

VVtt

tt

qq

HV SupplyHV SupplyHV SupplyResistorResistor

Air-CoreAirAir--CoreCoreTest ObjectTest ObjectTest Object

HV Switch

HV HV SwitchSwitch Voltage + PD

DetectionVoltage + PD Voltage + PD

DetectionDetection

Control and data processing unitControl and data Control and data processing unitprocessing unit

VVtt

tt

qq

HV SupplyHV SupplyHV SupplyResistorResistor

Air-CoreAirAir--CoreCoreTest ObjectTest ObjectTest Object

HV Switch

HV HV SwitchSwitch Voltage + PD

DetectionVoltage + PD Voltage + PD

DetectionDetection

Control and data processing unitControl and data Control and data processing unitprocessing unit

Fig. 2. - Schematic view of a 250kV DAC energizing diagnostics, Oscillating Wave Test System High Voltage.

3

cable sample (represented as a capacitance Cc) is linearly charged with a HV power supply in a few seconds to the selected test voltage level. As soon as the cable is charged, the HV supply is disconnected and a specially designed solid-state switch connects the cable sample to an air-core inductor in a closure time of less than 1µs. In this way, and LC loop is created and an oscillating voltage wave is applied to test the sample. Due to application to the cable section of a continuously increasing HV voltage supply, directly followed by a switching and oscillation period no ‘‘steady state’’ conditions occur.

The test frequency of the oscillating voltage wave is approximately the resonant frequency of the circuit and it is dependent on the cable capacitance. Due to the low loss factor and design of the air-core, the resonant frequency is close to the range of power frequency of the service voltage. The maximum power cable capacitance which can be tested at DAC stresses using OWTS HV system can be calculated in dependence of maximum voltage applied (figure 3). The energizing system as well as the PD measuring circuit practical solution as proposed by OWTS HV system consists of a number of components (figures 6 and 7). To load the power cable capacitance up to 250kV HV voltage supply has been used with a circuit effective load current of 8 mA. The function of the 250kV switch is to establish a series resonance circuit between charge HV power cable capacitance and the air inductor L. As a result damped AC stresses may

occur in the cable sample. The system inductor consists of five in series connected air coils with a total inductance of 4H. To make it sure that the damping of the voltage waves is mainly depending on the test object losses the inductor is air core type. Also the epoxy insulation of the windings provides PD freedom of this part. Due to the fact that for the above described technique: • the frequency of damped AC voltages is in the range of

power frequency of acceptable HV test systems,

Dielectric losses during DAC voltages

PD during DAC voltages

Time →

Voltage ↑

Dielectric losses during DAC voltages

PD during DAC voltages

Time →

Voltage ↑

Fig. 5.- Voltage withstand diagnosis using DAC voltages. During selected time at constant or increasing voltage application PD activity and dielectric losses can be measured.

Inductor L

HV Divider

250kV solid-state switch

250kV HV Supply

1600

2000

Resistor R

Control and data

processing unit

Inductor L

HV Divider

250kV solid-state switch

250kV HV Supply

1600

2000

Resistor R

Control and data

processing unit

Fig. 6. - Schematic over-view of the OWTS HV diagnosis system: total weight 300kg; occupied space 1.6m x 3m x 2m.

Dam

ped

AC

voltag

e

Partial discharges: PDIV, PD level, PD location

Dielectric losses

Dam

ped

AC

voltag

e

Partial discharges: PDIV, PD level, PD location

Dielectric losses

Fig. 4.- DAC voltage and diagnostic parameters which can be obtained during a measurement.

Fig. 3. - Dependence of the maximum power cable capacitance and the applied test voltage.

250kV HV Switch3.5 H Inductor

250kV HV Supply

150kV HV Power Cable

HV Control UnitPD Analyzer

250kV Divider

250kV HV Switch3.5 H Inductor

250kV HV Supply

150kV HV Power Cable

HV Control UnitPD Analyzer

250kV Divider

250kV HV Switch3.5 H Inductor

250kV HV Supply

150kV HV Power Cable

HV Control UnitPD Analyzer

250kV Divider Fig. 7. – PD testing of a 150kV XLPE power cable. In foreground the complete installation of OWTS 250 kV.

4

• a number of power cycles is applied to the cable sample provide ignition of PD sources in similar ways as compared to operating conditions,

the PD activity can be on-site measured with multiple undisturbed sinusoidal voltage cycles. For the purpose of PD localization by traveling waves, the bandwidth of the system amplifier is increased up to 10 MHz combined with a 100 MHz

digitizer. The PD activity signals, ignited during one or more oscillating voltage waves, are detected by the system, which can process the signals for two purposes. (1) A phase-resolved PD

pattern can be resolved from multiple DAC sequences. In this way, patterns can be obtained which are similar to those recognized under 50(60) Hz conditions (figure 8). (2) Single PD pulses can be analyzed for original location by using traveling wave analysis. Statistical evaluation of PD signals obtained after several oscillating waves can be used to evaluate the location of discharge sites in the power cable. A PD mapping is created, which shows the distribution of the detected PD in a cable circuit, as a function of the magnitude or the intensity (figure 9). In figure 10 example of PD diagnosis is shown.

B. Return Voltage Measurement It is known that the electric field stress on a dielectric causes two types of reaction: conduction and electrical polarization. Whereas current conduction is continuous movement of charge caries the electrical polarization is the alignments of dipoles in a material in the direction of electrical field with a limited charge displacement. As a result the measurement of return voltage (RVM) is related to polarization processes in dielectric under the test. Due to the fact that all aging processes change the way a dielectric responds to the application of an electric field the shape of the return voltages depends on the active polarization phenomena, figure 11. Using the system as shown in figure 12 the cable is charged with 1-2kV by means of a DC for 15 min. Then it is discharged via an internal discharge resistor softly for 2 sec. Afterwards the return voltage increases during 900-1800s measuring time. The shape of the return voltage curves correlates with the polarization effects. This process is repeated two times at different voltages levels.

PD vs Ut

0

2000

4000

6000

8000

10000

12000

14000

0.5 0.7 1 1.3 1.5 1.7

Meetspanning Uo

Ont

ladi

ngsn

ivea

u (p

C)

Fase L1

Fase L2

Fase L3

PD vs Ut

0

2000

4000

6000

8000

10000

12000

14000

0.5 0.7 1 1.3 1.5 1.7

Meetspanning Uo

Ont

ladi

ngsn

ivea

u (p

C)

Fase L1

Fase L2

Fase L3

Fig. 10. – PD evaluation of 50kV, 6km long power cable. From the PD analysis phase blue shows high PD activity as compared to other phases and the reference norms. Using PD mapping analysis of the phase blue the PD source has been located in the near cable termination.

Fig. 8. - Examples of DAC voltages as applied to a 220kV power cable showing PD activity in a termination; PD activity at 50kV (upper), PD activity at 100kV (middle), PD activity at 150kV (lower).

Fig. 9. - Example PD location mapping as made after DAC voltage stresses up to 190kV. In the graph the PD mapping shows the discharge concentration in the cable termination a function of the cable length. In the database table below measured PD quantities are automatically extracted for particular cable parts and accessories.

5

Using the system as shown in figure 11 the cable is charged with 1-2kV by means of a DC for 15 min. Then it is discharged via an internal discharge resistor softly for 2 sec. Afterwards the return voltage increases during 900-1800s measuring time. The shape of the return voltage curves correlates with the polarization effects. This process is repeated two times at different voltages levels. To evaluate the RVM measuring data two parameters are used in the CDS system see figure 12: (1) The ration Qa of the absolute return voltages as measured at 1kV and 2 kV during 15 minutes provides good indication of any decrease in the insulation

resistance. (2) To analyze the characteristic of the RVM voltage shape parameter p has been introduced. From PILC insulation is known that the factor increases with the ageing of the cable insulation. If p value is in a range below 0,2 the PILC insulation seems to be dry and in good condition.

IV. KNOWLEDGE RULES As shown in figures 4 and 5 the OWTS 250 system can provide information about dielectric losses in the test object. Using DAC voltages the dielectric losses can be derived from the decay characteristics of the oscillating voltage wave, voltage decay. In this way the dielectric losses as measured at different voltage levels can be used to evaluate the insulation condition of power cables. In figure 13 an example is shown of dielectric losses and RVM diagnostics as applied to an external-gas-pressurized 150kV mass-insulated power cable. After no PD activity has been measured up to 1.5U0, during voltage withstand diagnosis at increasing voltage level 1.0U0 up to 1.5U0 it has been observed that the dielectric losses in phase L3 are changing slightly as compared to phase L1 and L2. Due to the fact that this type of cable insulation has to be PD-free the information about the dielectric losses will be of importance for condition assessment of this cable section. To support the decision process of Asset Management (see table 1) based on experiences as obtained on the basis of field experiences the knowledge rules has been developed, figure 14. It follows from this figure that for cables in Cat A, B, C and D where over-all degradation is evaluated Asset Management information about availability in service can be determined. For a cable in category D, due to the presence of local insulation defect as detected by PD diagnosis it is not possible generate any information about the availability.

V. CONCLUSIONS In this paper new solutions for advanced condition assessment of HV power cables have been shown. New technology to generate on-site test voltages as suitable to detect, locate partial discharges and to determine dielectric losses is described. The use of RVM method for HV power cables is demonstrated. Based on experiences as collected till now the following can be concluded: I. Condition assessment of HV cables consists of

RVM shape factor

L1: 0.126 L2: 0.136

RVM shape factor

L3: 0.22

RVM Measurement Dielectric losses @ 0.8…1.5U0Phase: L1, L2

Phase: L3

RVM shape factor

L1: 0.126 L2: 0.136

RVM shape factor

L3: 0.22

RVM Measurement Dielectric losses @ 0.8…1.5U0Phase: L1, L2

Phase: L3

Fig. 13. - Examples integral diagnosis of a gas-pressurized, mass insulated 150kV cable, 6,2km.

Fig. 12. – Return voltage analysis: Voltage shape quantification by p, quotient of initial slopes by Qa

RVM measuring principles:

Forming phase: 900 sec

Discharging phase: 2 sec.

Measuring phase: 900-1800 sec.

RVM measurement (CDS System) on a 150kV power cable.

RVM measuring principles:

Forming phase: 900 sec

Discharging phase: 2 sec.

Measuring phase: 900-1800 sec.

RVM measurement (CDS System) on a 150kV power cable.

Fig. 11. – Return voltage measurement: measuring principles, CDS system during diagnosis of a 150kV gas-pressurized mass-impregnated power cable.

6

- Insulation weak-spots detection and localization: partial discharge occurrence in cable insulation and in cable accessories;

- Integral insulation condition estimation: all aging processes change the way a dielectric responds to the application of an electric field.

II. Based on technologies as available; OWTS 250 and CDS non-destructive diagnosis of above mentioned processes is possible;

III. However it is not possible to determine the remaining life of a cable section, based on field experiences knowledge rules can be determined to support the maintenance decision processes.

Since more than one year these methods are successfully used in the field and therefore in the next time intensive and systematic field measurements on power cables in European transmission grid will be performed to develop knowledge rules for condition assessment of HV power cable systems cables.

VI. REFERENCES [1] E. Gulski, F.J. Wester, W. Boone, N. van Schaik, E.F. Steennis,

E.R.S. Groot, J. Pellis, B.J. Grotenhuis, ‘’Knowledge Rules Support for CBM of Power Cable Circuits’’ Cigre Paris 2002, SC 15 paper 104

[2] E. Gulski, F.J. Wester, Ph. Wester, E.R.S. Groot, J. W. van Doeland, ‘’Condition Assessment of High Voltage Power Cables. ’ Cigre Paris 2004, SC D1 paper D1.306

[3] W. Hauschild, W. Schufft, R. Plath, K. Polster, "The Technique of AC On-Site Testing of HV Cables by Frequency-Tuned Resonant Test Systems", CIGRE 2002, paper 33-304.

[4] E. Gulski, F.J. Wester, J.J. Smit, P.N. Seitz and M. Turner “Advanced PD diagnostic of MV power cable system using oscillating wave test system”, IEEE Electrical Insulation Magazine, 16, 2, 2000, p. 17-25

VII. BIOGRAPHIES Dr.hab.ir. Edward Gulski was born in 1958 in Poland. In 1982 he received from Dresden University of Technology in Germany the M.S. degree in information technology. From 1982 until 1986 he worked as research assistant at the HV laboratory of Dresden University of Technology. In 1987 he joined the HV laboratory of the Delft University in the Netherlands where he

performed research in the field of partial discharge diagnostics. In 1991 he received his PhD degree from Delft University of Technology. In 2004 he received his Doctor Habilitatus degree from Warsaw University of Technology. At present, he is associate professor involved in education and research in the field of insulation diagnosis of HV components and Asset Management. He is member of the Executive Board of KSANDR organisation and responsible for research in education. He is chairmen of Cigre Working Group G D1.33.03 “PD measurements’’ and chairmen of Cigre Working Group D1.17 ''HV asset condition assessment tools, data quality, and expert systems'' and Member of IEEE Working Group 400.3.

Eng. Paul P. Seitz, born 1966 in the United States, Stoneham Mass. In 1991 he received his Master degree in Electrical Engineering from the ABB Engineering School. In 1998 he received his Diploma at the Youth Business Management School HWV Olten. In 1993 he founded together with father Paul N. Seitz the Seitz Instruments AG Switzerland. Seitz Instruments AG is today a major vendor of PD-Diagnosis-Systems worldwide. Main experience field:

Software & Hardware development of PD-Analysis-Systems & High-Voltage Measurement Systems. Currently he is CEO of Seitz Instruments AG, Switzerland.

Dr.Ing. Frank Petzold was born 1955 in Dresden (Germany). From 1975 to 1979 he studied electrical engineering at the High Voltage Institute of the Technical University Dresden. In 1984 he received his PhD degree from the High Voltage Institute of Technical University Dresden with the Dissertation ”Influence of several stabilizers on tree-inception in XLPE insulations” From 1984 to 1990 he was chief of the development department in the cable factory KWO Schwerin (Germany). In the time from 1990 to 1999 he worked as self employer with his own consulting company for cable fault location and leak detection. Since 2000 he is Technical Director of SEBA Dynatronic GmbH Germany, responsible for R&D and technical marketing. Mr. Petzold is Member of several working groups in IEC; CIGRE and IEEE.

Eng. Edwin R.S. Groot MBA was born in Renkum in the Netherlands, on October 13, 1968. He graduated from the electro technical Highschool in Alkmaar and studied at the Academy for management on the University of Groningen and he graduated for his Master degree on the University Nyenrode.

He researched the technical and organisational impact of condition based maintenance for high

voltage equipment and he also made an international research about the impact of privatisation and liberalisation of the energy market. From 1995 mr. Groot was intensive involved in the introduction and implementation of condition based maintenance in the Netherlands. Nowadays in his capacity as manager within the Dutch utility Nuon and as managing director of the international organisation Ksandr he is still involved in the maintenance and refurbishment projects for high voltage infrastructure.

Off-line method interpretation rules for combination of tan δ and PD diagnosis to assess and classify the condition of oil filled power cables

< 3-5 years> 434hrs-<95%Cable replacement

tan δ > 80PD >5000 + no localization

E

????-<50%Inspection/repair PD location site

tan δ ≤ 80PD > 1000 + localization

D

> 15 years< 434hrs<72hrs>95%Follow trend

tan δ > 50 tan δ ≤ 80

PD ≤ 1000 or PD ≤ 5000 + no localization

C

> 25 years<87,7hrs<48hrs> 99%no≤ 5 0PD ≤ 1000B

> 30 years< 8,7hrs<24hrs> 99,9%no≤ 30PD ≤ 250A

Expected life time

Unplanned maintenance

/year

Planned maintenance

/year

Availability in service

Advisetan δ at U0[x10-4]

PD at U0 [pC]CAT.

Availability = 1 – [Unplanned maintenance/ ( 365 x 24hrs) – CBM/year]

Off-line method interpretation rules for combination of tan δ and PD diagnosis to assess and classify the condition of oil filled power cables

< 3-5 years> 434hrs-<95%Cable replacement

tan δ > 80PD >5000 + no localization

E

????-<50%Inspection/repair PD location site

tan δ ≤ 80PD > 1000 + localization

D

> 15 years< 434hrs<72hrs>95%Follow trend

tan δ > 50 tan δ ≤ 80

PD ≤ 1000 or PD ≤ 5000 + no localization

C

> 25 years<87,7hrs<48hrs> 99%no≤ 5 0PD ≤ 1000B

> 30 years< 8,7hrs<24hrs> 99,9%no≤ 30PD ≤ 250A

Expected life time

Unplanned maintenance

/year

Planned maintenance

/year

Availability in service

Advisetan δ at U0[x10-4]

PD at U0 [pC]CAT.

Availability = 1 – [Unplanned maintenance/ ( 365 x 24hrs) – CBM/year]

Fig. 14. – Knowledge rules as developed for oil-filled HV power cables.