PD TEST PROCEDURE AS AFTER LAYING ACCEPTANCE TEST ON

HV AND EHV POLYMERIC CABLE SYSTEMS L. Fornasari

Techimp H.Q. SpA Zola Predosa, Bologna, Italy

e-mail: lfornasari@techimp.com

G.C. Montanari, A. Cavallini Department of Electrical, Electronic and Information

Engineering - DEI University of Bologna

Bologna, Italy

Abstract— During the last two decades the use of XLPE polymeric cables for High Voltage and Extra High Voltage transmission lines has grown significantly, due to increased electrical congestion of metropolitan areas and the proven reliability of XLPE power transmission cables.

High quality extrusion processes and cable materials, along with more stringent factory acceptance tests, are at the basis of this increased reliability. Extruded cable drums, as well as cable accessories, are PD tested in the factory in order to check for the presence of insulation defects. However, when cable systems are installed on site, jointing is one of the most important steps of the overall laying process to achieve a high reliability. At this stage, small defects can be introduced, causing PD activity at operating voltage once in service, and leading to failure in relatively short times. For this reason, the best practice to obtain an effective after-laying commissioning test is to perform both HVAC test and PD test. PD test can identify smaller defects which can be missed by HVAC withstand tests.

This paper discusses the main differences between different test procedures (e.g. simultaneous and sequential) and test voltages, which are used to check the installed cable system condition. Furthermore, general considerations on the requirements that a PD testing device needs to fulfill are presented, along with two working cases where an improved PD system with separation and identification tools was able to find PD activities which were not highlighted during the HVAC test.

Voltages of 1.2-1.4 U0 seems to be effective to assess insulation condition during HVAC test. If an effective technology is used, both simultaneous and sequential PD tests are able to spotlight minor defects in XLPE cable insulation system.

Keywords-component; High Voltage; Partial Discharges; Ploymeric; Cables; T-F Map.

I. INTRODUCTION At the end of manufacturing and during the life of any

electrical (E)HV cable system, tests are performed in order to assess the insulation system condition. Keeping apart quality control and diagnostic tests during cable operation, this paper will focus on on-site after laying acceptance tests. These tests are performed at the installation site, in order to check if transport and assembly have caused any critical defect, which may lower the dielectric withstand voltage of the insulation

below an acceptable value. Dielectric withstand tests (DWT) plays a fundamental role

in assuring quality of electrical equipment, thanks to its clear go/no-go acceptance criteria. IEC62067 Par.16.3 (AC voltage test of the insulation) [1] establishes DWT as mandatory for HV and EHV cables. Testing waveform should be substantially sinusoidal, with frequency between 20 Hz and 300 Hz. A voltage according either to TABLE I or to 1.7 U0 (depending on practical operational conditions) shall be applied for 1 hour. The suggested UTEST values were determined considering the technical limitations of testing equipment, usually Resonant Test Sets (RTS). The value 1.7 U0 (≈ √3 ) is the maximum voltage at which the cable can be energized, generally achieved using a high voltage transformer with delta-star connection.

TABLE I. VOLTAGES USED FOR AC WITHSTAND INSULATION TEST

CABLE RATED VOLTAGE

NOMINAL U0

(FOR UTEST DETERMINATION)

SUGGESTED UTEST

RATIO UTEST/U0

220 to 230 kV 127 kV 180 kV 1.39 275 to 287 kV 160 kV 210 kV 1.32 330 to 345 kV 190 kV 250 kV 1.26 380 to 400 kV 220 kV 260 kV 1.13

500 kV 290 kV 320 kV 1.11

Alternatively a voltage of U0 can be applied for 24 hours directly from the network, and this is the so-called “soak test”. As few constrains are given, HVAC test procedures and voltage levels are agreed normally between the purchaser and the contractor. Note that the standard does not mention the PD test as a useful test for commissioning service, even if the importance of such test is worldwide known. Being this standard so general, many questions are left open:

• Which voltage is most appropriate for HVAC test? • Which PD test procedure, sensor layout and PD

instrument characteristics are most suitable to carry out a successful after laying test?

• Is PD test really useful for commissioning tests, or is HVAC test sufficient to assess insulation condition state?

This paper will address the above issues, trying to give clear and unambiguous answers.

523

2013 Electrical Insulation Conference, Ottowa, Onterio, Canada, 2 to 5 June 2013

978-978-1-4673-4744-0/13/$31.00 ©2013 IEEE

II. EFFECT OF HVAC TEST VOLTAGE Voltage levels during HVAC test should be carefully

chosen, keeping in mind the test purpose. The main purpose of the test (not considering PD measurement for the moment), is aiming at highlighting major defects in the insulation system, which would lead to failure before the end of the design life. Sometimes, independently of cable system voltage level, final users follow “the higher the better” strategy, asking for HVAC test at 1.7 U0, an approach that is fully compliant with IEC62067. However, testing at too high voltage level may cause break down of the cable due to defects that would never be activated under whole design life under nominal working conditions. Alternatively, it could create a permanent damage which is able to trigger PD under nominal operating conditions, bringing the cable to premature failure. Therefore, testing at much higher voltages than the maximum sustained overvoltage the cable can experience during its life should be considered carefully.

Sustained power-frequency overvoltages are the most harmful voltage-related phenomenon for high voltage systems, since they can trigger PD activity which can be still active when normal operation condition is restored. Overvoltages due to insulation fault occur on a three-phase network when the neutral is unearthed or impedance-earthed. More precisely, when an insulation fault occurs on phase A, an overvoltage earth fault factor, K, is defined as (in p.u.) [2].

3 · 1|2 | (1)

(2)

where and are the zero-sequence and direct reactance of the network evaluated at the fault point. If the neutral is completely unearthed, is infinite, and K equals to √3. In practice, ranges from 1 for completely earthed neutral, to 3. Using 3, it can be concluded that 1.25.

Ferranti effect must be taken into account also when a power cable is connected at the end of a very long overhead line. In case the load is suddenly disconnected (cable side), resonance effects take place and the result is a voltage increase. However, in most cases voltage increase is not too high: for a 500 kV overhead line it can be at worst 1.16 U0.

Therefore, the choice of a proper maximum voltage to be applied to the cable for both HVAC and PD tests is a critical issue. Too low voltage levels may not help increasing in-service reliability, while too high voltage levels will require extra RTS costs, or may damage unnecessarily the insulation system.

As a compromise between costs and benefits, the best solution is trying to find out the proper maximum voltage test on the basis of network characteristics. If the goal is to assess that the system is able to withstand sustained overvoltages, voltage levels of 1.2-1.4 U0 could be already enough.

III. SELECTION OF PROPER PD SENSORS FOR HVAC CABLES The core point of an effective after laying test which

includes PD measurements is the choice of proper PD sensors to maximize measurement sensitivity, selected on the basis of the kind of accessories under test. Different typologies of PD sensors can be used for HV and EHV Cable applications, e.g. [3,4]:

• Capacitive sensors: can be embedded in accessories. • Flexible Magnetic Couplers (FMC): sensors wrapped

around the cable, which are sensitive to high frequency magnetic field of PD pulses.

• High Frequency Current Transformers (HFCT): inductive sensors sensing PD pulses through ground or sheath connections.

• Ultra High Frequency (UHF) antenna: antennas used for PD detection in the UHF range.

In order to obtain the Phase Resolved PD Pattern (PRPD), it is necessary to synchronize PD signals with respect to the voltage applied to the cable under test. For this purpose, Rogowsky coils can be used, placed around the cable itself or around the screen connection. PD Sensor choice and location issues are discussed in the following.

Outdoor Termination In case a shield surge arrester is present, it is suggested to

bypass it using a jumper cable connecting the termination metallic screen directly to ground. PD measurements are still possible (although with lower sensitivity) even with the shield surge arrester connected, as it will behave as a high-pass filter. A drawing of PD sensor installation is reported in Figure 1. Suitable PD sensors are:

• Embedded Capacitive sensor, if it is present, it is the most sensitive PD measurement point. Generally a quadrupole is connected to its output, allowing both PD and synchronization signals to be obtained.

• HFCT on the ground lead of the termination. This sensor can always be used since HV and EHV terminations are grounded.

• FMC wrapped around the cable close to the termination is a good sensor which can also help for PD detection.

GIS Termination In case surge arresters are connected across the insulating

ring, it is suggested to remove them temporarily and install a jumper cable linking the upper part of the GIS with the lower metallic body of the termination. A drawing of PD sensor installation is reported in Figure 1. Suitable PD sensors are:

• Embedded Capacitive sensor, as in outdoor termination. • HFCT on the ground lead of the termination. • HFCT on the jumper cable across insulating ring.

This solution offers higher sensitivity with respect of HFCT around termination ground lead.

• FMC wrapped around the cable close to the termination. • UHF antenna placed on the insulating ring, which can

detect electromagnetic emission coming from PD activity in both stress-cone region and GIS region close to the termination.

524

Figure 1: PD Sensor’s connection configuration for Outdoor

Termination (left) and GIS Terminations (right)

Sectionalized joint During cable operation, sheaths are cross-bonded at

sectionalized joints. In order to perform off-line PD measurements, sheaths should not be cross-bonded. Therefore, prior testing, it is necessary to remove the cross-bonding bars and install jumper cables connected between the inner and the outer conductors of each coaxial cable coming from joints. After this, one more jumper cable per phase has to be installed between the first jumper and ground (totally 2 jumpers per phase). Applying the above screen configuration modification, a “straight” to ground connection will be present in all sectionalized joints.

Figure 2: PD Sensor’s connection configuration for Sectionalized joints:

a) Accessible Joint, b) Joint Link Box, c) Joint PD Box

A drawing of PD sensor installation is reported in Figure 2. Suitable PD sensors are:

• Embedded Capacitive sensor generally connected to an accessible PD Box through proper adapters by means of coaxial cables. It is possible to use a quadrupole, allowing both PD and synchronization signals to be achieved.

• HFCT sensor placed directly around screen connections if the joint is accessible

• HFCT around jumper cable inside the link box, if the joint is not accessible.

• FMC wrapped around the cable close to the joint, if joint is accessible.

IV. SELECTION OF PD TESTING EQUIPMENT Choosing the proper PD equipment is fundamental for

effective after laying diagnosis in HV and EHV cables. Off-line calibration procedure, and consequent PD measurement output in pC, can be performed only when HV conductor and screen are accessible. The calibration pulse has to be injected, for each accessory to be calibrated, between the HV conductor and the closest ground. In a long cable circuit this holds for terminations, but not for joints. A proper calibration cannot be performed at cable joints, because the HV conductor is not accessible. In addition, using the calibration constant obtained at terminations for joints PD detection has little meaning, as attenuation cannot be taken into account. Furthermore, in order to perform proper PD measurements in pC, the IEC60270 [5] bandwidth (< 500 kHz) should be used or at least included in the bandwidth of the detection system (in case of Wide-Band PD detectors). According to CIGRE 502 document from WG D1.33 [6], it is not recommended to use detectors working only in the IEC bandwidth for on-site tests, due to the strong background noise. Thus, conventional PD measurements are often abandoned in favor of non-conventional ones performed using UWB systems. In this case, the suitable bandwidth is up to several tens of MHz, allowing better SNR and complete waveform detection. For this reason, on-site PD tests are normally performed in mV, and the calibration is replaced by a sensitivity check, which is able to spotlight insufficient sensitivity. In order to provide an apparent charge estimate for a joint, a test should be performed preliminarily in laboratory with the same kind of accessory, deriving approximate calibration coefficient.

Due to the above issues, PD detectors commercially available with a bandwidth of 16 kHz – 35 MHz will completely fit sensitivity needs. An additional advantage is that the complete waveform of each PD or noise pulse could be detected, allowing noise rejection and PD location through advanced algorithms.

Noise-rejection is by far the most serious issue to be solved. An effective tool is the so-called T-F map [7-8-9]. It provides two parameters for each recorded pulse: the equivalent time, T, and equivalent bandwidth, F (frequency-domain information). By this way, it is possible to achieve complex PD measurements, tracking different phenomena individually and assigning different warning levels to them, depending on their harmfulness. Each cluster on the T-F map includes all pulses having the same waveform characteristics: PD pulses with similar waveforms are supposed to come from the same source (defect). Therefore different clouds means different phenomena.

An example of PD acquisition on a 220 kV joint using HFCT inside the link box is reported in Figure 3. As it can be seen, an enhanced denoising and separation technique was able to reveal the existence of a small interface PD activity inside the joint itself (qMAX95%=3.4 mV) hidden by disturbance. This defect did not lead to breakdown during the 1 hour HVAC test.

525

The last step after the choice of proper PD sensors and detection unit, is the selection of PD test procedures. Two different approaches are presented in the following: simultaneous test during HVAC test and sequential PD test.

Figure 3: 220kV Joint. Example of Separation and rejection of external

disturbances allowed by T-F map. PD activity was below the noise level.

V. SIMULTANEOUS AND SEQUENTIAL PD TESTING

A. Simultaneous PD test during HVAC test When testing one phase at a time, there are a few

prerequisites to consider for carrying out proper simultaneous PD tests:

• PD equipment and sensors needed equal to the number of accessories

• Power supply needed at each accessory site or PD equipment fed from internal battery

• Central control unit (CCU) for PD equipment control • Fiber optic ring laid between each detection point and

CCU. Before testing, an on-site installation activity for

predisposition PD detectors connections to each accessory must be planned. PD acquisition parameters must be adjusted on the basis of background noise. When feasible, voltage should be applied by steps in order to record the Partial Discharge Inception Voltage (PDIV), and kept for 1 hour at the maximum test value, UTEST. During the HVAC test, PD measurements should be performed continuously and autonomously by PD acquisition units, but acquisition parameters should be checked manually and regularly by operators from the CCU. This is essential, because noise changes during the test, particularly when the cable is energized. After the 1-hour HVAC, voltage is decreased following the same voltage steps, and in case any PD has been detected, PD Extinction Voltage (PDEV) is also recorded. PD data have to be carefully analyzed for each acquisition unit to complete the PD test report.

An example of a simultaneous PD test during HVAC test of a 220kV XLPE utility cable circuit is shown in the following. During voltage rise-up at 100 kV an internal PD activity was

found coming from the outdoor termination (opposite side with respect of RTS). The utility decided to continue anyway the HVAC test. Termination failed after some minutes at 180 kV phase-to-ground. PD test results together with a picture of the defective stress cone are depicted in Figure 4.

Figure 4: PD measurements carried out during HVAC test. Internal PD detected it the 220 kV termination.

B. Sequential PD test after HVAC test The other common after-laying test approach is sequential

PD test following the HVAC test. The aim of this test is to highlight possible small defects which have been missed during the High Voltage test. Sequential PD tests need a number of PD equipment equal to the number of testing operators. The test is usually performed one phase at a time, through the following steps:

• N operators install PD equipment at N accessories; • Voltage is applied in three voltage steps: below U0, U0

and maximum agreed voltage for PD test (normally 1.2 – 1.4 U0) in order to record possible PDIV. During each voltage step, which is usually lasting about 10 minutes, each operator performs PD measurements on the assigned accessory. If PD is detected, PDEV is also evaluated.

• Voltage is shut down and operators move to the next N accessories to be tested. These steps are repeated until PD measurements are performed at all accessories.

An example of PD detected during a sequential PD test is shown in the following. The circuit was 800 meters long with one outdoor termination, one joint bay and one GIS termination, and passed the HVAC test at 1.4 U0. Strong disturbances due to bad RTS HV connections were detected, but despite that, it was possible to detect an internal PD activity using both joint capacitive sensor and HFCT in the link box. Separation through T-F map was effective and PD (of amplitude much smaller than noise) was located in the cable itself by means of the TDR technique, roughly 30 m from the tested joint, outdoor termination side. This PD activity was observed at all voltage steps i.e. 0.8-1.0-1.2 U0, so that it was decided to replace the defective segment with a new one, by adding a joint. PD test acquisition through joint embedded sensor and its processing is reported in Figure 5.

526

Figure 5: 220 kV XLPE Cable. PD test using Joint embedded sensor at 1.2 U0, Internal PD detected in cable insulation. Sequential procedure.

The defect inside the cable was not observed by factory quality control (it was created possibly by mechanical strain during laying), nor the HVAC test was able to highlight it. This underlines the importance of PD detection as the last critical step of HV cable commissioning.

C. Comparison between Simultaneous and Sequential PD procedures

A summary of a comparison between simultaneous and sequential PD tests, highlighting relevant advantages and disadvantages in TABLE II.

TABLE II. SIMULTANEOUS AND SEQUENTIAL PD TESTS

PROS/CONS OF SIMULTANEOUS SEQUENTIAL SENSITIVITY TO

SPORADIC PD ACTIVITY HIGH MEDIUM

PD LOCATION POSSIBLE, CONSTRAINS OF HVAC TEST

POSSIBLE, EXTRA TIME IF NEEDED

PD DETECTED BEFORE BREAKDOWN DURING

HVAC TEST POSSIBLE NOT POSSIBLE

TIME CONSTRAINS FOR PD TEST YES (1HOUR) NO

RELIABILITY OF PD ACQUISITION HIGH HIGHER THAN

SIMULTANEOUS

CABLE STRESS 1 VOLTAGE APPLICATION

N VOLTAGE APPLICATIONS

PROBLEMS TO PD INSTRUMENT OR

COMMUNICATION

VERY SERIOUS PROBLEM AS TEST

CANNOT BE STOPPED

NOT A SERIOUS PROBLEM

FIBER OPTIC RING NECESSARY NOT NECESSARY RTS COSTS LOWER HIGHER

PD TEST COSTS HIGHER LOWER

As can be seen there are advantages and disadvantages in both approaches. For simultaneous tests, the time constraint of 1 hour can be too tight in case of difficult data detection and interpretation, and an high reliability of PD acquisition units and fiber optic network is required to ensure that the test is successfully performed. Sequential tests are almost immune to these problems and do not need any fiber optic ring, but they may require longer test times.

VI. CONCLUSIONS The above considerations emphasize how HVAC test and

PD test play a fundamental role in HV and EHV cable on-site

commissioning. Voltage levels for HVAC test should be carefully selected. While 24 hours soak test is suggested only when no other options are available, a too high test voltage may initiate defects which can lead to premature failure under service conditions. Voltages of 1.2-1.4 U0 seems to be effective to assess insulation condition, considering the typical magnitudes of sustained overvoltages in HV networks.

PD diagnosis is of outmost importance. Procedures for both simultaneous and sequential testing have been discussed and analysed. From a diagnostic point of view (considering the same maximum test voltage) there is no significant difference between those two approaches. In case a fiber optic ring is laid already, the simultaneous approach is applicable and can help saving test time and cost. Where a fiber optic ring is not available, sequential test could be the best solution.

It should be pointed out that power cables ageing is not due to electric field only. Mayor faults during in-service condition are due to thermal or mechanical factors, as thermal cycles or uneven mechanical stresses in accessories. These phenomena may cause delaminations, triggering PD activity and leading to unexpected failures.

For this reason it is always suggested to install continuous on-line PD monitoring system, in order to detect possible defects related to operating conditions which are not active during cable commissioning [10]. Such system allows correlation of PD activity to other operating parameters such as load, temperature, overvoltages etc., enabling smart alarm algorithms based on T-F map separation and trending over time [11]. In this way, a complete understanding of power cable ageing is possible, avoiding in-service failures and scheduling maintenance outage only when really needed.

VII. REFERENCES [1] IEC 62067, High Voltage Power Cables “Power cables with extruded

insulation and their accessories for rated voltages above 150kV (Um=170 kV) up to 500 kV (Um=550 kV) – Test Methods and requirements”, 2nd edition, 2011.

[2] D. Fulchiron, “Overvoltages and insulation coordination in MV and HV”, Cahier Technique Merlin Gerin n°151, pp.3-5, 1995.

[3] G. C. Montanari “On Line Partial Discharge Diagnosis of Power Cables”, IEEE EIC, Montreal, QC, Canada, 31 May - 3 June 2009.

[4] IEEE Standard 400.3, “IEEE Guide for Partial discharge Testing of Shielded Power Cable Systems in a Field Environment”, 2006.

[5] IEC 60270, High Voltage Test Techniques, “Partial discharge Measurement”, 3rd edition, March 2001.

[6] Cigre Brochure 502, “High-Voltage On-site Testing with Partial Discharge Measurement”, WG D1.33, June 2012.

[7] A. Contin, A. Cavallini, G.C. Montanari, G. Pasini, F. Puletti, “Digital detection and fuzzy classification of partial discharge signals”, IEEE Transactions on Dielectr. Electr. Insul., Vol. 9, No. 3, pp. 335-348, 2002.

[8] A. Cavallini, G. C. Montanari, A. Contin, and F. Puletti, “Advanced PD Inference in On-Field Measurements. Part I: Noise Rejection”, with IEEE Trans. on. Electr. Insul., Vol.10. N.2, pp. 23-30, April 2003.

[9] M. Tozzi, G. C. Montanari, A. Cavallini, “PD detection limits in extruded power cables through wide and ultra-wide bandwidth detectors”, IEEE Trans. Dielectr. Electr. Insul., Vol. 15, No. 4, pp. 1183-1189, 2008.

[10] L. Fornasari, A. Cavallini, G.C. Montanari, “Advanced Condition Monitoring of insulation systems: a building block for Smarter Grids”, Conference Record of the 2012 IEEE Condition Monitoring and Diagnosis conference (CMD), pp.533-537, Bali (Indonesia), 23-27 September 2012.

[11] L. Fornasari, G. C. Montanari, A. Cavallini, "Alarm management in permanent PD monitoring for generators" Conference Record of the 2012 IEEE International Symposium on Electrical Insulation (ISEI), pp.571-575, San Juan (PR), 10-13 June 2012.

527