CIRED

20th International Conference on Electricity Distribution

Prague, 8-11 June 2009 Paper 0923

CHARACTERIZATION OF PARTIAL DISCHARGES BEHAVIOR IN XLPE DISTRIBUTION CABLE JOINTS

Quintn CORRIONERO, Luis SANTOS Hidrocantbrico Distribucin Elctrica SAU, Spain qcorrionero@hcenergia.com, lsantos@hcenerga.com Jos COTO, Manuel COTO, Ruben LPEZ Universidad de Oviedo. Spain jcoto@uniovi.es Miguel ARNIELLA HC Explotacin de Redes, SAU, Spain LuisMiguel.ArniellaCano@hcenergia.com

ABSTRACT
This paper discusses research conducted to characterize medium voltage power cables PD activity which will be used to develop an on-line monitoring system. Data of the PD characteristics that take place in different models of joints performed on 20 kV XLPE cables are presented. Furthermore, a measurement system that enables the on-line partial-discharge detection and location based on Rogowski coils is discussed and designed.

while minimizing or eliminating some potential adverse charging effects of the direct voltage high-potential test method. In the other hand, an on-line monitoring system is being development based on Rogowski coils and high speed digital acquisitions units.

TEST ARRANGEMENTS
Laboratory performance test of power cable joints with monitoring of PD activity has been developed using off-line VLF test, according to VDE DIN 0276-620/621 (CENELEC HD 620/621). The equipment uses a test frequency with a 0.05 Hz to 1 Hz range to energize the cable up to a voltage level of 20 kV. This voltage range will allow to generate the partial discharge activities in the cable without causing dangerous electric stresses on the dielectric materials. The well-know reflectometry technique is adopted in this tests. The discharges are captured and measured by the measurement unit in which the difference in the travelling time between the first incoming pulse and its reflection is used to compute the locations of partial discharge along the cable. [1]. Nevertheless, practical application of PD measuring methods on medium-voltage cable systems, and particularly VLF test methods, is almost restricted to off-line diagnostics, requiring the cable to be taken out of service for the measurement period. A system capable of measuring PDs while the cable is in-service, i.e. on-line, provides considerable advantages. First, the power supply is not discontinued during the measurement and therefore the cost of on-line measurement should appreciably be lower than off-line measurement. There is no need for the heavy and expensive test voltage supply and the coupling capacitors are often replaced with inductive couplers, current transducers, placed around cable. Also, the cable system is tested under real operating conditions allowing continuous monitoring, so time-related information is detected. The laboratory measurements were conducted in the PD Laboratory of HC ENERGIA using the measurement configuration presented in Fig. 1.

INTRODUCTION
The DENISE research project is led by the Spanish electric utility ENDESA and a consortium of over 19 firms and research centers, developing various technologies related to the concept of the intelligent distribution networks. The activities developed by HC ENERGIA and the Electrical Engineering Department of the University of Oviedo corresponds with task 5.3 Predictive Maintenance. The significance of high reliability of distribution cables increases at the time of higher competition between utilities and due to regulators requirements. Many problems of underground cable systems are caused by internal defects of the joints. Faults arise from wrong assembly of accessories, voids or impurities, etc... Joints are generally known to be the bottleneck of underground distribution lines. The present work, in order to realize the possible defects at the cable joints interfaces, studies the joints behavior based on the partial discharge activity measurement. Partial discharge (PD) measurement is well known as a nondestructive method to investigate defects in the power cable system. Using partial discharge measurement with source localization, direct allocation of PD activity on joints, cable segments or cable terminations is enabled. Laboratory performance test of power cable joints with monitoring of partial discharge has been developed using VLF test and on-line monitoring. The experiments were performed at HC ENERGIA test facility. VLF testing techniques, effective for a broad range of cable types, provide a method of evaluation, and help to fill the need for more complete information on the cable system condition

CIRED2009 Session 1

Paper No 0923

CIRED

20th International Conference on Electricity Distribution

Prague, 8-11 June 2009 Paper 0923

Fig. 3. Rogowski coil equivalent circuit. The geometric characteristics of Rogowski coil are given in table I. Fig. 1. Test setup used for measuring PD activity. The VLF equipment was a Baur PD Portable system, which consists of PD level measurement unit and a PD source localization unit, and a BAUR VLF sine wave generator. In order to develop the on-line system, the measuring set-up was also performed by means of a signal transducer and a digital data-acquisition system. The PD activities are recorded in a digital oscilloscope and subsequently transferred to computer software in order to implement the signals processing tasks. The proposed on-line method is based on a simple and low cost device: a Rogowski coil, which feed a Yokogawa DL750 SCOPECORDER, an oscilloscope and data recorder unit with 12 bit A/D-converter. Table I. Geometric characteristics of Rogowski coil. Dimensions Value Inner diameter a 0.0409 m Outer diameter b 0.0715 m Wide W 0.0648 m Diameter of wire d 0.00055 m Length total of wire Lcu 1.76848 Number of turns of coil N 12 The lumped parameters can be calculated as follows [3, 4]:

Rc = c
Lc =

Lcu d 2

(1)

0 N 2W b ln 2 a
4 2 0 (b + a) b +a ln ba 4 2 0 (b + a) b +a ln ba

(2) (3)

DESIGN OF THE ROGOWSKI COIL

The Rogowski coil operates on based the theory of electromagnetic coupling. The necessity of the constructed Rogowski coil is to measure partial discharged, which are pulses with frequency of several tens of Mhz. [2]. Fig. 2 and 3 show the construction of the Rogowski coil and equivalent circuit diagram.

Cc =

Cc =

(4)

Where Rc, Lc, Cc and M are the resistance, inductance, capacitance and mutual inductance of coil. 0, 0, and 0 are the copper resistivity, air permeability and permittivity, respectively. The theoretical and real parameters of Rogowski coils are given in table II. These parameters were experimentally obtained through an Agilent 42194A impedance analyzer. Table II. Rogowski coil parameters Parameter Theoretical Real Rc (Ohm.) 0.116120659 0.124615 Lc (H) 1.04242E-06 1.89E-06 Cc (F) 1.5092E-11 1.0088E-11 M (H) 8.68684E-08 NA

Fig. 2. Rogowski coil dimensions.

CIRED2009 Session 1

Paper No 0923

CIRED

20th International Conference on Electricity Distribution

Prague, 8-11 June 2009 Paper 0923

To characterize the performance of the joints and evaluate the Rogowski coil performance, a set of six VLF tests were conducted. The analysis started at 5 kV with 1 kV increments, up to 20 kV. A summary of PD activity, registered by the Baur equipment, is shown in tables III-IV. On these tables, some differences between moisturepolluted joints and gas-pollutes joints can be seen: Fig. 4. Frequency response of Rogowski coil. Fig. 4 show the bode diagram (amplitude-frequency and phase-frequency response). This bode diagram is calculated with a value of impedance Zout =100 . The moisture shows less activity of PD than the air for service voltage (12 kV) because of the difference of dielectric between water and air. When the voltage is higher than service voltage, the moisture exceeds in activity in comparison with the air. The level of discharges of air is always higher than the moisture because of the air permittivity is much less, thereby, the level of critical inception voltage must be higher to produce the dielectric breakdown in air.

TEST RESULTS
PD signals are recorded from a 320 m. long, 12/20 kV 1x95 Al XLPE cable. Two joint are assembled in this probe: a polluted joint and an undamaged joint, which served as a control sample. Two types of contaminated joints were tested: moisture-polluted joints and gas-polluted joints. The cable joints were assembled using kits and cables supplied by HC ENERGIA. Joints and cables were representative of components used in the HC distribution system. Table III. PD Activity. Per cent of discharges
5 -10 kV Damp polluted joint Gas polluted joint (air bubbles) Polluted joint Cable end Control joint Polluted joint Cable end Control joint 0% 100% 0% 0% 100% 0% VLF Test Voltage 10-12 kV >12 kV 12.59% 87.41% 0% 42.75% 56.49% 43.39% 59.01% 35.14% 5.86% 50.66% 43.39% 43.39%

Otherwise, with regard to the Rogowski coil behaviour, the set of test were sampled at 10 MS/s. The PD data captured by the Rogowski coil is essentially an oscillatory voltage pulse, which needs to be processed to obtain PD characteristics. The voltage pulse captured by the Rogowski coil sensor for a 50 nC calibrator pulse is shown in Fig. 5. With a propagation velocity of the total cable system about 160 m/s, the difference in time of arrival for the direct and reflected pulses are 4 s.

PD Activity (% of discharges)

Fig. 5. Rogowski coil response for 50nC calibrator pulse. The PD activity at the end cable using VLF test is shown in Fig. 6. Fig. 7 and 9 shows the PD activity captured by the Rogowski coil from de cable end and the damaged joint respectively. Sources of noise in this signal, largely filtered through the coil, are generally AM radio, FM radio and mobile communication signals in air with frequency bands of 144-30000 kHz, 88-108 MHz, 890-1880 MHz, respectively [5]. As seen this noised signal partially hidden the PD activity that might be present. Discrete Wavelet Transform DWT-based denoising methods [5, 6, 7] can be used to extract PD pulses from noised signals. Fig. 8 and 10 shows the denoised signals where the time interval between pulses confirm the PD source at cable end and damaged joint (50 m. from de far cable end).

Table IV. PD Activity. Level of discharges

5 -10 kV PD Activity (level of discharges, pC) Damp polluted joint Polluted joint Cable end Control joint Polluted joint Cable end Control joint 0 31.22 0 0 35.34 0 VLF Test Voltage 10-12 kV >12 kV 583.1 1648.6 0 1009.2 1887.6 538 620 1897 856 1229 1969 952

Gas polluted joint (air bubbles)

CIRED2009 Session 1

Paper No 0923

CIRED

20th International Conference on Electricity Distribution

Prague, 8-11 June 2009 Paper 0923

difference of behaviour between moisture-pollutes joints and gas-polluted joints. A Rogowski coil has been designed and tuned and as it has been demonstrated, provides a high sensitivity, good linearity, enough wide band as a probe to measure and locate partial discharges. Further tests will be conducted to characterize damaged joints and improve the Rogowski coil response. Redesign of the experimental setup has begun to improve the on-line monitoring system. Future plans include the design of new Rogowski coils and the use of higher sampling frequencies.

Fig. 6. PD signal from the cable end. VLF response.

REFERENCES [1] V. Dubickas, 2006, On-line time domain reflectometry diagnostics of medium voltage XLPE power cables, Licentiate Thesis, Royal Institute of Technology, Stockholm, Sweden 2006. [2] J. Zhu, L. Yang, J. Jia and Q. Zhang, 2005, The Design of Rogowski Coil with Wide Band Using for Partial Discharge Measurements, Proceedings of 2005 lntemational Symposium on Electrical Insulating Materials, Kitakyushu, Japan. [3] M. Argueso, G. Robles and J. Sanz, 2007, Identification of parameters in a Rogowski coil used for the measurement of partial discharges, IEEE Instrumentation and Measurement Technology Conference. IMTC, Warsaw, Poland, 2007. [4] G. Murtaza, 2008, Partial Discharge Detection For Condition Monitoring Of Covered-Conductor Overhead Distribution Networks Using Rogowski Coil Doctoral Dissertation. Faculty of Electronics. Helsinki University of Technology. ISBN 978-951-229447-3. Fig. 9. PD signal from the damaged joint. [5] Zhou, X. Zhou, C. Kemp, I.J., 2005, An improved methodology for application of wavelet transform to partial discharge measurement denoising, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 12, No. 3, 586- 594. [6] Shim, I.; Soragan, J.J.; Siew, W.H.; Sludden, K.; Gale, P.F., 2000, Robust partial discharge measurement in MV cable networks using discrete wavelet transforms, IEEE Power Engineering Society Winter Meeting, Vol. 1, 718 - 723. [7] Ma, X.; Zhou, C.; Kemp, I.J., 2002, Interpretation of Wavelet Analysis and Its Application in Partial Discharge Detection, IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 9, No. 3, 446 - 457.

Fig. 7. PD signal from the cable end. Rogowski coil response.

Fig. 8. Rogowski coil response after denoising.

Fig. 10. PD signal from the damaged joint after denoising.

CONCLUSIONS
This is preliminary research to characterize incipient failure behavior in underground distribution cable. A first round result obtained data from experiments of laboratory is the

CIRED2009 Session 1

Paper No 0923