A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

DIELECTRIC LOSS CHARACTERIZATION OF HYDRO-QUEBEC MEDIUM VOLTAGE (MV) SUBMARINE CABLES - PHASE 1

Jean-François DRAPEAU, IREQ (Hydro-Québec), Varennes, Québec, Canada, drapeau.jean-francois@ireq.ca

Bruno CANTIN, Hydro-Québec Distribution, Québec, Québec, Canada, cantin.bruno@hydro.qc.ca

Nigel HAMPTON, Joshua PERKEL, NEETRAC, Atlanta, GA, USA, nigel.hampton@neetrac.gatech.edu, joshua.perkel@neetrac.gatech.edu (3) ABSTRACT

This paper presents the results of a study carried out in order to perform a condition assessment of seven MV submarine cables operated by Hydro-Québec Distribution (HQD). The assessment was based of the principle of dielectric loss characterization. Two methods were chosen: Time Domain Spectroscopy (TDS) and VLF Tan delta. Results show that combining the two methods provides a valuable source of information. The use of VLF Tan delta also allows for comparison with the extensive measurement database of tests performed on other equivalent cable systems.

KEYWORDS

Submarine Cables, Dielectric Loss, Very Low Frequency (VLF), Tan Delta, Time Domain Spectroscopy (TDS)

INTRODUCTION

Hydro-Québec Distribution (HQD) operates several MV submarine cable links in order to supply customers living on Île-d’Orléans, Île-aux-Grues and Isle-aux-Coudres islands along the St-Lawrence River. These links constitute critical assets in the HQD underground system, as they are the only source of power for the residents of these islands and they represent a significant capital investment. Furthermore some of these submarine cables are approaching or exceeding the typical “design life” of these types of cable systems, with ages ranging from 29 to 48 years.

Prior to 2007, none of these cable systems had been subjected to any kind of diagnostic testing. At this time a test program was initiated to obtain numeric re-interpretable data in order to assess the health of these cable systems. All the submarine cable systems in this study are protected against water ingress. Those insulated with XLPE and EPR are equipped with lead sheaths and are mechanically protected with steel wire armor screen. Paper cables are “pipe-type” design, so their protection is intrinsically provided by the steel pipes. However, most of them are simply laid on the river bed. Therefore they are subject to salt water tidal currents and they do not benefit from the protection of burial. Hence, one of the concerns for HQD relates to the possible occurrence of any insulation degradation that could be associated with local or global water ingress. Also, in order to allow for accurate planning of refurbishment or system upgrades, it is of interest to be able to monitor the evolution of the health (or condition) of the insulation with time.

In order to address these issues, a diagnostic testing program was initiated. The objective of this program was to measure and eventually monitor over time the global insulation condition of a set of selected submarine cables, using dielectric spectroscopy. For this purpose, dielectric loss was characterized, for a set of voltages up to

operating voltage (U0).

Phase 1 of this project consisted of performing an initial condition assessment for each cable. The plan for phase 2 will include the acquisition of trending information from subsequent measurements made at regular intervals (ex. 3 to 5 years). This paper presents the results for phase 1. A preliminary condition assessment will also be discussed.

SPECIAL ISSUES WITH SUBMARINE CABLES

Compared to those installed in traditional cable systems (URD, feeders, etc.), submarine cables present a number of special issues.

Length: In most cases, submarine cable segments are very long compared to those of traditional underground land cable systems. Instead of one to several hundreds of meters, submarine cables lengths tend to be in the range of one to few thousands meters.

Architecture: In addition to being characterized by much greater lengths, submarine cable systems have typically very few or zero joints.

Value: Installation costs of submarine cables are typically much higher than that of traditional land cable systems: generally at least several times more expensive on a per unit length basis.

Criticality: In addition to the cost, installation of submarine cables requires much longer times than land cables: the cable supply needs to be planned well in advance and the installation process is far more complex and takes longer time. Also, in many instances, utilities cannot afford to operate on more than one cable. Hence, reliability is a key issue since there is typically no backup. The criticality issue has a great impact on testing strategies: diagnostic techniques and procedures are expected to be selected in a way that minimizes the risk of failure on test.

Effort: Given the value and criticality of submarine cir-cuits, submarine cable operators naturally tend to be more willing to invest in testing efforts in order to maximise the amount of information that could be obtained on the condition of the cable system while favouring the best achievable accuracy. Concerns would be leaning more on the “input” issue compared to land cable situations where the attention is more typically focused on the “throughput”.

DESCRIPTION OF SUBMARINE CABLES TESTED

The study presented in this paper covers 7 submarine cable circuits, each consisting of three-phase (3ϕ) medium voltage (MV) cables operated at 25 kV, including:

Close and Return

A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

• Three XLPE based insulation (3ϕ) cables (1973, to mid 1980s)

• One EPR based insulation (3ϕ) cable (1997)

• Three paper based insulation (3ϕ) cables (1962, to 1981)

Details on these cable circuits are provided in Table 1. The two longest submarine cables are 5.9 km and 5.2 km long. The shortest are the three submarine cables

supplying Île-d’Orléans, near Québec City: 1.2 km each. Figure 1 shows the geographic layout of the 7 cables along St-Lawrence River. As mentioned earlier, all the submarine cables are exposed to salt water and tidal currents. Île-aux-Coudres is located in a section of the St-Lawrence River where tidal currents are known to be especially strong. Accordingly, the cables were built with protection against water ingress that includes a lead sheath and with mechanical protection provided by a steel wire armor screen (see Figure 2).

Table 1: Description of the submarine cables tested

LOCATION CIRCUIT / SUBSTATION

COMMISS. YEAR

TYPE

LEVEL

CONDUCTOR

SIZE LENGTH

(KM) CAPACITANCE

(UF) OTHER

INFORMATIONS

Île-aux-Coudres PAU-234 1997 EPR 25 kV 750 kcmils 5.2 1.80 4 land joints

Île-aux-Coudres PAU-234 R 1973 XLPE 25 kV 1/0 3.3 0.49

Isle-aux-Grues TAC-235 1986 XLPE 25 kV 1/0 5.5 0.79

Isle-aux-Grues TAC-235 R 1986 XLPE 25 kV 1/0 5.9 0.85

Île-d’Orléans Ange-Gardien 1962 Pipe-type(paper) 25 kV 4/0 1.2 0.32

Île-d’Orléans Ange-Gardien 1973 Pipe-type(paper) 69 kV 400 kcmils 1.2 0.39

Île-d’Orléans Ange-Gardien 1981 Pipe-type(paper)

69 kV 400 kcmils 1.2 0.43

Figure 1: Geographical layout for submarine cables at Hydro-Québec Distribution

INSULATION

10 km

5 mi

Île-aux-Coudres 2 cables XLPE & EPR-based

Isle-aux-Grues 2 cables XLPE-based

Île-d’Orléans 3 cables Paper-based

Ref. for satellite view: GoogleMapsTM.ca

Close and Return

A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

Figure 2: Construction of XLPE submarine cable installed at Île-aux-Coudres

DIELECTRIC LOSS MEASUREMENTS

In this study, the condition assessment is based of dielectric loss measurements. Two methods were used: Time Domain Spectroscopy (TDS) and dielectric loss measured under Very Low Frequency (VLF) AC voltage (otherwise known as VLF Tan delta). Both methods rely on the principle of applying a high voltage to the test sample while monitoring the dielectric loss. However, considering the “criticality” issue of submarine cables described earlier, a decision was made never to exceed the service voltage under any circumstance.

Time Domain Spectroscopy (TDS): TDS is an off-line, non destructive and efficient method for measuring dielectric losses of many types of insulation. The principle is shown in Figure 3. The dielectric loss is calculated in the low frequency range (covering a spectrum between 10-1 and 10-4 Hz) from the polarization (resulting from the application of a DC voltage) and depolarization currents (measured while cable circuit is grounded through an electrometer). The protocol for voltage application with time is shown in Figure 4. The tests were carried out using a TDS device developed at IREQ [1] (Figure 5).

Figure 3: TDS polarization - depolarization sequence & Hamon approximation

Figure 4: TDS time-voltage application protocol

Figure 5: TDS device developed at IREQ

VLF Tan Delta Measurements Tan δ values are obtained by applying an AC voltage and measuring the phase difference between the voltage waveform and the resulting current waveform. This phase angle is then used to resolve the total current (I) into its charging (IC) and loss (IR) components. The Tan δ is the ratio of the loss current to the charging current.

Figure 6 shows an ideal equivalent circuit for a cable, consisting of a parallel connected capacitance (C) and a voltage dependent resistance (R). The Tan δ measured, at a frequency ƒ and voltage V, is the ratio of the resistive (IR) and the capacitive (IC) currents. Figure 7 shows the setup used in this study for performing VLF Tan δ evaluations.

V

I

RI CI

VRI

ICI

δ

θ

Figure 6: Equivalent circuit for Tan δ measurement and phasor diagram

As compared to TDS, the VLF procedure provides dielectric loss information at one frequency at a time. The frequency value that is the most commonly used is 0.1 Hz. TDS is able to obtain this information for multiple frequencies simultaneously (approx. range: 0.05 Hz to 0.0005 Hz), hence the concept of “spectroscopy”.

Close and Return

A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

Figure 7: Test setup for VLF submarine cable condition evaluation

Some recent work [2] has shown that when TDS and VLF dielectric loss measurements are brought together, the overall loss spectroscopy is generally consistent. Thus, the VLF Tan δ data at 0.1 Hz can be used to extend the loss spectrum obtained using the TDS. In order to maximize the information content from the current test program, and to obtain further confirmation of the consistency of the TDS and VLF loss on the frequency spectrum for various types of insulation, decision was made to perform “VLF spectroscopy” by using VLF frequencies of 0.1 Hz and 0.02 Hz, time permitting.

Table 2 shows the comparative durations of TDS and VLF protocols.

Table 2: Typical durations of TDS and VLF test protocols

TEST

PROCEDURE TIME DETAILS OVERALL

TIME (MIN)

TDS For each voltage step: POL: 200 s ; DEPOL: 500 s 3 voltage steps (one phase)

35

VLF (0.1Hz)

For each voltage step: 1st step: 3 min, then 2 min for all the others 3 voltage steps (one phase)

7

VLF (0.02Hz)

For each voltage step: 1st step: 12 min, then 10 min for all the others 3 voltage steps (one phase)

32

RESULTS AND INTERPRETATION

Table 3 shows the results from VLF Tan δ testing, described according to the following VLF diagnostic features [3]: mean Tan δ (at 0.5 U0 and 1 U0), differential Tan δ (difference between Tan δ at 1 U0 minus Tan δ at 0.5 U0), also called “Tip-Up”, and time stability (measured by the standard deviation of individual Tan δ values at 1 U0).

Figures 8 through 12 show the combined loss spectrum obtained from TDS and VLF testing for XLPE, EPR and paper insulations, respectively.

Table 3: VLF testing results °0.1 Hz

CABLE SAMPLE INSULATION

TYPE

0.5 U0

(E-3)

1 U0

(E-3)

TIP-UP 1U0-0.5U0

(E-3)

TIME

STABILITY

STD DEV

(E-3)

PAU-234 Ph.A EPR 2.01 2.00 -0.01 0.009

PAU-234 Ph.B EPR 1.96 1.97 0.01 0.007

PAU-234 Ph.C EPR 1.99 1.99 0.00 0.009

PAU-234-R Ph.A XLPE 0.15 0.16 0.01 0.004

PAU-234-R Ph.B XLPE 0.16 0.16 0.00 0.007

PAU-234-R Ph.C XLPE 0.17 0.18 0.01 0.010

TAC-235 Ph.A XLPE 0.07 0.09 0.02 0.007

TAC-235 Ph.B XLPE 0.10 0.10 0.00 0.007

TAC-235 Ph.C XLPE 0.11 0.14 0.03 0.009

TAC-235-R Ph.A XLPE 0.09 0.10 0.01 0.011

TAC-235-R Ph.B XLPE 0.09 0.10 0.01 0.010

TAC-235-R Ph.C XLPE 0.10 0.11 0.01 0.007

Ange-G(1962) Ph.A Paper 5.85 4.44 -1.41 0.012

Ange-G(1962) Ph.B Paper 7.01 5.19 -1.82 0.011

Ange-G(1962) Ph.C Paper 6.51 4.77 -1.74 0.007

Ange-G(1973) Ph.A Paper 4.10 3.78 -0.32 0.011

Ange-G(1973) Ph.B Paper 3.24 3.03 -0.21 0.010

Ange-G(1973) Ph.C Paper 3.77 3.48 -0.29 0.009

Ange-G(1981) Ph.A Paper 2.55 2.48 -0.07 0.010

Ange-G(1981) Ph.B Paper 2.57 2.50 -0.07 0.010

Ange-G(1981) Ph.C Paper 2.55 2.47 -0.08 0.009

Figure 8: Typical loss spectrum for XLPE insulated submarine cable system

Figure 9: Typical loss spectrum for EPR insulated submarine cable system

Submarine cable

VLF TAN δ

Close and Return

A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

Figure 10: Loss spectrum for a 1962 paper insulated cable system

Figure 11: Loss spectrum for a 1973 paper insulated cable system

Figure 12: Loss spectrum for a 1981 paper insulated cable system (same design that 1973)

In the cases of EPR and paper insulations, the TDS polarization and VLF loss spectrums are quite consistent with each other. Extrapolation of the TDS loss spectrums to higher frequencies generally fits well with the VLF Tan δ spectrums. In the case of XLPE, the apparent mismatch shown in Figure 8 has also been reported in previous work [2] and could be attributed to sensitivity limitations of the measuring devices, since related Tan δ levels are in a pretty low range (< 0.3 X 10-3 for VLF and polarization DC currents < 80 nA).

It is interesting to note that the EPR and paper loss spectrums maintain their consistency with voltage dependence. Figure 10 provides a good example where an ”inverse dependance” of loss with voltage (”negative” Tip-Up, i.e. Tan δ value decreases with increasing voltage), typical for paper insulation, can be observed for VLF as well as for TDS polarization.

Figure 13 presents a summary of dielectric loss measurements, combining TDS polarization Tan δ values at 0.001 Hz with VLF Tan δ values at 0.1 Hz. Each graph symbol is related to dielectric loss measurements for one submarine cable circuit. This allows for a phase to phase comparison. Under normal ageing circumstances, one should expect the dielectric loss between phases not to differ significantly.

Figure 13 Summary of dielectric loss measurements

The blue straight dotted line in Figure 13 corresponds to a constant ratio of TDS over VLF losses. It is parallel to the black one which corresponds to a ratio of 1 (i.e. TDS = VLF). It can be noticed that all TDS loss values are higher than their corresponding VLF values. Accordingly, the blue line is shifted upward. Such a way to interpret the results brings to the fore one interesting feature about dielectric spectroscopy, which is the information regarding the magnitude of the frequency dependance of Tan δ: this magnitude translates into the loss frequency spectrum as the value of the slope: ΔTan δ/Δf. The examination of that feature could reveal some valuable information about the nature and the condition of the insulation system. As for example, healthy XLPE cable typically show Tan δ values with a rather ”flat slope” (i.e. Tan δ stays rather constant with frequency). On the other hand, EPR and paper insulation show a moderate decrease in loss with frequency (i.e. low negative slope). High slope values can provide an indication of some weakness in the insulation system, especially for the case of PE-based insulation.

For the submarine cables with paper-based insulation considered in this study, Figure 13 shows that the Ange-Gardien(1973) cable circuit has the largest variation between phases and the greatest difference between TDS and VLF loss measurements. Even though this is an interesting observation, there is not enough field evidence with this behavior to make a condition assessment based on this information alone. These features might be worth examining during the next phase of this study.

PRELIMINARY CONDITION ASSESSMENT

The primary feature upon which the condition assessments for these submarine cables is based is temporal trending. As discussed above, a second round of dielectric loss measurements is planned to be made on these circuits 3 to 5 years after this first round. But for the time being, the data can be put into context, thanks to the

Close and Return

A.7.6 8th International Conference on Insulated Power Cables A.7.6

Jicable’11 – 19 – 23 June 2011, Versailles - France

extensive collation of VLF dielectric loss data [3] carried out as part of the Cable Diagnostic Focused Initiative (CDFI) by Neetrac. Using this database, it is possible to get a ”preliminary” appreciation of the condition of the submarine cables considered in the HQD study. Figure 14 shows the cumulative distributions of Tan δ values at U0 measured in the US on cable systems with cable lengths equal or greater than 1500 m (i.e. 5000 ft).

1001010.10.01

100

75

50

25

0

Tan Delta (E-3)

Pe

rce

nt

Lengths > 5000 ft

Figure 14: VLF Tan δ cumulative distribution for circuits with cable length > 5000 ft

According to this figure, Tan δ values relating to the submarine cables in this study are:

• within the lowest 1% of Tan δ values recorded in the database for XLPE and EPR-based insulations

• within the lowest 5% of Tan δ values recorded in the database for paper-based insulation.

CONCLUSIONS

This paper presents results from the first phase of a study undertaken with the objective of obtaining the most reliable information regarding the condition assessment of HQD submarine cables.

It is important to note that compared to traditional land cable systems, submarine cables present a number of special issues such as long length, architecture with few or no joint, high value, high criticality and test strategy.

The condition assessment parameters selected for the purpose of this study are based on dielectric loss measurements. Two methods have been chosen: Time Domain Spectroscopy (TDS), previously developed at IREQ, and VLF Tan δ.

Results show that combining the two methods provides a valuable source of information: TDS provides additional diagnostic information from dielectric spectroscopy while VLF Tan δ allows the results of this study to be related to the knowledge base that has been developed with this technology.

A comparison of the current VLF data for the submarine cables concerned in this study to the values that are part of an extensive collation of VLF data reveals that the actual HQD submarine cables VLF Tan δ values stand in the lowest 1% of Tan δ values recorded for XLPE and EPR-based insulations and in the lowest 5% of Tan δ values recorded for paper-based insulation.

FUTURE WORK

In the near future, HQD plans to conduct a more extensive investigation of the condition of the three paper submarine cable systems that are supplying Île-d’Orléans. This investigation will include VLF PD assessment and OWTS (Oscillating Wave test System) PD and dielectric loss measurements.

In the longer term, HQD plans to perform a second round of dielectric loss measurements within the next five years to acquire trending data.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the technical contributions of Mr Daniel Jean, technician at IREQ, for making the test equipment fit to the field conditions. We would like also to extend our thanks to Mr. Yves Magnan, engineer at HQD who played in important role in the initiation of this project, and to the HQD crew personal who provided assistance in the field. Finally, the authors would like to mention the invaluable support of the US Department of Energy to the CDFI project (under award number DE-FC02-04CH11237), as well as that of the numerous participants and supporters of the CDFI.

REFERENCES

[1] J.F. Drapeau et al, 2007, “Time Domain Spectro-scopy (TDS) As A Diagnostic Tool For MV XLPE Under-ground Lines,” JICABLE07, Versailles, France.

[2] J.F. Drapeau and J.C. Hernandez, 2008, “Measure-ment of Cable System Losses using Time Domain and VLF Techniques”, Presentation to Sub F, IEEE Insulated Conductors Committee (ICC), San Antonio, TX.

[3] J. Perkel et al, 2011, “Challenges Associated with the Interpretation of Dielectric Loss Data from Power cable System Measurements,” JICABLE11, Versailles, France.

GLOSSARY

HQD: Hydro-Québec Distribution

TDS: Time Domain Spectroscopy

VLF: Very Low Frequency

PD: Partial Discharge

OWTS: Oscillating Wave Test System

XLPE EPR

Paper

0.1 1 10 100 1000

Tan Delta (E-3)

Close and Return