OPTIMAL PARTIAL DISCHARGE DIAGNOSIS OF MEDIUM VOLTAGE POWER CABLES

E. Gulski1*, E. J. Sinambela 1, P. Cichecki1, J.J. Smit1, F. Petzold2, F. de Vries3 and P.P. Seitz4 1 Delft University of Technology, 2628 CD Delft, Mekelweg 4, the Netherlands 2 SebaKMT, 96148 Baunach, Dr. Herbert Iann Str. 6, Germany 3 Alliander, 1817 DD Alkmaar, Voltastraat 2, the Netherlands 4 Seitz Instruments AG, 5443, Niederrohrdorf, Mellingerstrasse 12, Switzerland *Email: e.gulski@tudelft.nl Abstract: Sensitive insulation condition assessment of service aged medium voltage power cables is one of important tasks in maintenance of power networks. In particular by off-line energizing of the cable circuits using external voltage sources may provide sensitive on-site partial discharges (PD) detection. An obtaining of a good picture of the actual insulation condition by means of PD diagnosis depends on several aspects. Many parameters can influence the quality of the measurement and several issues are important during the execution of a PD measurement. In this paper, several parameters which can influence the quality of the measurement and issues important during PD measurements and analysis will be discussed. In order to optimalize the testing procedures and to provide an effective use of on-site PD diagnosis for condition assessment of distribution power cables guideline procedures are proposed.

1.

INTRODUCTION

Reliability and availability of a medium voltage (MV) power network is determined by the condition of all components in that power network [1]. Due to the fact that more than half of the breakdowns in the cable network are caused by internal fault in the insulation systems or accessories of the cable network, diagnostics of new installed (after-laying test) and service aged distribution power cable are very important to prevent such breakdowns and get knowledge about actual condition of particular system. By knowing the condition of the cable, the early action can be done before the breakdown occur during operation. The most accepted on-site diagnostics of MV power cables are based on off-line energizing of a cable section using continuous or damped AC voltages and standardized PD detection and analysis, figure 1 [2, 3]. It is known that a) due to physical and stochastic character of the PD phenomena in function of defect type and its electrical characteristics e.g. inception conditions, PD level in function of electric stresses on the one hand and b) the complexity of the test objects like power cables e.g. different length, different insulation materials, different accessories, local electromagnetic interferences on the other hand, the PD detection and analysis are complex. In order to minimize the problems in performing PD measurement and to provide an optimal use on-site PD diagnosis will be proposed in this paper covering the following procedures: 1. Test object definition: most relevant information about the test object parameters e.g. insulation type cable length, number, type and the position of the joints is essential in the execution of all

Figure 1: Example of on-site PD testing of 13km 10kV XLPE insulated cable (tested in section) using damped AC voltages. 2. further steps, Calibration procedure: to obtain correct measuring results application of standardized measurement is relevant, PD measurement test procedure: test duration and the height of the test voltage are important to obtain most significant information e.g. about the PD occurrence, PD data collection and analysis: the amount of diagnostic data, the way of data analysis are relevant to obtain a good picture about PD activity in the particular cable section, Condition assessment: the transition from measuring into insulation condition information is based on key-data extraction and interpretation. Finally only this information is used in further decisions about maintenance, operation and/or replacement.

3.

4.

5.

2.

TEST OBJECT DEFINITION

For MV power cable networks is characteristic, that there is a large variety on cable insulation and accessories types, switchgear types and the way how the data e.g. the length, number and position of the joints of a particular cable section are available at the utility. An optimal execution of the field test which consists of several steps like connection of the test system, the test procedure, data collection and finally the test results interpretation depends from the quality of the information about the test object. Table 1: Basic elements of test object information
Cable System Cable section identification Operating voltage Type of construction Installation year Name of cable manufacturer Cable insulation Conductor type and size Cable length. Installation year Type of accessories Name of cable manufacturer Location of the accessories Installation year Test date, time / Inspector / Comment
v avg = 2 l2 t 2

Insulation Cable

Accessories

Inspection Information

Missing information about the length of the cable section or about the exact number and the position of joints may influence the quality of the test data and can result in misrepresentation and falsification of test results. E.g. wrong information about the cable length results in erroneous TDR calibration, which in its turn may result in incorrect location of PD sites in the cable. 3. CALIBRATION PRCEDURE

Figure 3: Generic procedure for standardized PD on-site calibration of power cables. generating sinusoidal, continuous or damped AC voltages and providing standardized PD detection. Performing off-line PD detection using conventional method is an important step in assessing the sensitivity of the particular circuit. The purpose of this step is to determine the value in pC of the smallest PD signal detectable under the test conditions. Performing standardized PD detection and localization on power cables diagnostics requests two calibration procedures: Calibration of the PD reading in [pC]; calibration of PD reading is performed according to [4]. This calibration is made by injecting defined current pulse to the termination of the cable system. A certain PD pulse (e.g. 50pC) is injected to the cable system and the reading of the PD detector is calibrated for the given circuit (including test object) parameters. Calibration of the PD pulse propagation velocity in [m/s] reading; to localize PD sites using time domain reflectometry (TDR) analysis the PD pulse propagation for different PD magnitudes has to be determined. In this calibration, PD pulse is injected to the cable system; original pulse and reflected pulse are detected by PD detector, figure 2. The propagation velocity of the pulse in a particular cable system is calculated by the time different between the original PD pulse which recorded in at the near end and the reflection pulse. As a result, standardized calibration is very essential because this is a single method to determine the PD reading of the PD detector and to made the performance check of the whole circuit. In order to support correct calibration that can be used in performing PD measurement, a procedure is proposed in figure 3.

The most accepted test method of on-site diagnosis of MV cable networks is the so-called standardized offline method [4]. For this purpose cable system has to be disconnected from the network. The off-line PD detection consists of an external voltage source [5],
(a)

(b)

Figure 2: PD pulse propagation examples [1]: (a) example of the calibration of the propagation velocity in a XLPE cable system, (b) example of the localisation of a PD source by analysing the PD wave time differences.

4.

PD MEASUREMENT AND TEST PROCEDURE PD background noise estimation

4.1.

The sensitivity of the PD measurement tends to decrease with the increasing of the environmental noise. Disturbances can occur even if the cable system is not energized. To achieve a good PD measurement, the knowledge about PD background noise is important. The PD background noise can be measured e.g. by performing PD measurement on the whole circuit at 0kV up to few kVs test voltage. This information is important to distinguish between the inception of PD related to defect of the test object and external PD caused by noise and disturbances. If background noises higher than acceptable level, the source of background noise should be found and eliminated. In the case of the background noise is originates from external sources and cannot be eliminated, this background noise is noted as reference in PD analysis (including noise suppression techniques).

inception of lower PD pulses will be not registered and a cable with PD of 180pC could wrongly be declared as PD-free. To avoid this false negative judgment it is recommended to use as low as possible measuring ranges, starting at the situation where the backgrounds noise level is e.g. below 50% and at increasing test voltage to observe carefully the changes in the PD pulse occurrence. Moreover, increase the test voltage up to e.g. 1.7U0 needs continuous control of the PD measurement sensitivity. Otherwise by selecting measuring range e.g. up to 200pC whereas the actual PD activity in function of the test is in the range of 2nC will result in wrong estimation of the maximum PD magnitude of 200pC. To avoid this false negative judgment it is recommended to use as low as possible measuring ranges, and in the case of increase of PD level to adopt the PD measurement sensitivity by avoiding effects of signal clipping. 4.3. Selection of the test voltage level

PDEV

PDIV

U0
PD-free

U0

U0
Not PD-free

Figure 4: Application of PD measurement in function of the AC over-voltage test. Any particular discharge larger than q0 (cable specific acceptance value) measured below selected test voltage Ut leads to rejection of the object under test, see shaded area. PD inception voltage PDIV and PD extinction voltage PDEV referred to U0 are important indicators of PD presence during network operation. 4.2. Selection of PD measurement sensitivity

The quality of PD measurement is strongly affected by selection of proper PD measurement sensitivity. This sensitivity depends on the PD detector measuring range as used to collect data. Most PD detectors are not able to measure real-time at the same time PD pulses in the ranges few pCs, hundreds of pCs and tens of nCs. Therefore the measuring range of the PD detector has to be selected in such a way that the PD range is not too high or not too low compare to the maximum PD magnitude expected at the PD inception voltage (PDIV) and during the increase of the test voltage. By selecting PD measuring range e.g. up to 1nC and using trigger threshold e.g. of 25% will trigger the PD detector on PD pulses larger than 250pC. As a result

It is known, that there is a large diversity on insulation defects which may occur not only in the accessories of a cable but also in the insulation. Moreover, it is know that PD inception in all these defects depends on the local electric field enhancement. As a result with regard to insulation defects different situations may occur in a cable system (insulation and/or accessories). It was already shown many years ago [6] that the best way of PD detection as well as the recognition of different discharging defects is related to changes of the applied electric field, see figure 4. This interaction between PD inception voltages (PDIV), PD extinction voltage (PDEV), and PD magnitude at different voltage stresses provides more detailed picture about the PD processes in different types of cable insulation defects. Obtaining this information in its turn is important in interpretation of discharging defects: 1. The level of PDIV as compared to the nominal service stress is of absolute relevance in determining the degradation ration; e.g. discharging defect with if PDIV = 0.3Uo is mostly more severe that a defect with PDIV =0.9Uo. 2. To declare a power cable PD-free the PDEV has to be at least 1.3Uo. 3. Analyzing the increase of PD level and PD intensity in function of the applied voltage e.g. between PDIV and 1.7Uo is a good indicator of defect severity. Summarizing it can be concluded that due to diversity of discharging defects and the complexity of cable systems for sensitive PD diagnosis of power cables the voltage dependence of PD processes is important. In order to support correct calibration that can be used in performing PD measurement, a procedure is proposed in figure 5. 5. PD DATA COLLECTION AND ANALYSIS

While performing PD measurement on distribution cable system, the existence of partial discharge above specified sensitivity is detected and relevant

(a)

(b)

Figure 7: Measured PD pulses in the time domain obtained from cable terminations (t = tcal) [1]: (a) PD event originating from the near termination, (b) PD event originating from the remote termination. Recognition which termination is concerned can be performed by pulse width analysis. PD pulse Y is wider than PD pulse X. 5.1. PD mapping

Figure 5: Generic procedure for standardized onsite PD measurement on power cables.

Figure 6: Generic procedure to obtain a PD finger print of a power cable. information about the PD occurrence can be called fingerprint of a particular power cable section. The fingerprint may consist of the following two types of data: PD mapping and PD parameters. In Figure 6 a procedure is proposed showing consecutive steps in obtaining the PD finger print of a power cable.

PD mapping is the representation of the PD occurrence in the length of the cable system. As a result the PD activity in the weak spots of the cable insulation or of the accessories e.g. terminations or splices can be shown. In particular, analyzing PD mapping in function of the test voltage as illustrated in figure 4 is useful to distinct between location of different discharging sites (defects) and their inception. Using conventional (standardized) PD diagnostic the PD mapping the PD mapping is performed by using TDR [7]. The location of PD activities is determined by calculating the time different between the original pulse and the reflection pulse. The accuracy of PD localization is determined by the accuracy in selecting of original and reflection pulse during calibration. The following are problems that might occur during selection of original and reflection pulse in TDR: Detection PD occurrence in the terminations: PD source originated from one of terminations is indicated by the time difference between original pulse and reflection pulse is equal to the delta calibration time (t =tcal). The problem may be faced in determining the position of the termination whether in remote termination or near termination. Therefore the position of the PD source can be determined by analyzing the width of the pulse. PD pulse originating from remote termination is wider than PD pulse originating from near termination (see figure 7). PD source is located closely the remote termination: When PD source is located closely to the remote termination, time different between the original and the reflection pulse is smaller than tcal and the detected PD pulses are superimposed. Figure 8(a) shows the PD pulses are superimposed in the time domain. Multiple PD sources in a cable system: Multiple PD events can be active in a cable system during

(a)

(b)
Figure 8: TDR examples of PD site location [1]: a) example of the PD location of a PD event located closely (44m) to the remote end of a 3-core PILC cable of 321m. The resulting detected PD pulses are superimposed, b) multiple PD events occurring in a time span of 45s only. Analysing the time difference between the incident pulses (Ai, Bi, Ci) and the reflected pulses (Ar, Br, Cr) shows that successive PD events A and C are from the same location and event B from a second location. performing PD measurement at the same time [1]. Figure 8(b) shows multiple PD events occur in a cable system analyzed using TDR. Analyzing time different between the original pulses (Ai, Bi, Ci) and their reflection shows that A and C have the same time different. It indicates that PD event A and PD event C are originated from the same location and PD event B is originated from another location. 5.2. PD parameters

In particular, based on TDR analysis, each PD pulse can be analyzed for its location in the cable system. A PD mapping of a cable system represents the distribution of all located PD events. Especially, the concentrations of PD events at one location represents a PD related defect, figure 9(a). It follows that the result of the PD location analysis for the different test voltage levels is shown for a 3-phase PILC cable system of 577m. PD concentrations are found in a joint (A), in a termination (C) and in a cable part (B). The cable accessories are indicated as black dots at the horizontal axis. The PD mapping can be shown for each individual phase or for all phases together. From the analyzed PD location mapping, the PD occurrence frequency as a function of the cable length can also be derived, see figure 9(b). The PD occurrence frequency in a location uses the total number of PD events inside a window of e.g. 1% of the total cable length (the general error in PD location) for that specific location. In the above example shows that the overall intensity, as obtained for all measurements at different voltage levels is the highest for the PD concentration at the remote termination, which is on the right corner of the plot. 6. CONDITION ASESMENT

The final result of PD measurement should be provided in the form of PD measurement report. Based on this report the condition assessment of the cable system can be determined. The PD measurement report should be able to describe the actual condition of the cable system so that the owner can used this report as data input for maintenance decision.
(a)

For the determination of the insulation condition, several PD parameters are found of importance and can be used to evaluate a PD measurement. For field application of the PD diagnostics, the PD properties as shown below turned out to be of relevance from practical experiences. The properties are measured directly by the detection equipment for the cable system, or derived after analysing the measurement data for individual cable components from the PD mapping. The PD properties are collected in the socalled fingerprint of a cable system, table 2. Table 2: Basic elements of test object information.
Cable system PD inception voltage PD extinction voltage PD magnitudes at ~V V: PDIV, 1Uo, 1.7Uo PD pattern PD intensity PD mapping PD inception voltage PD magnitude at ~V V: PDIV, 1Uo, 1.7Uo PD occurrence frequency Cable component

(b)

Figure 9: Example of PD location view of a 3phase PILC power cable of 577m [1]: (a) PD mapping, individual PD events as a function of the cable length, (different phases indicated by colour), (b) PD intensity plot, the concentrations (inside 1% of the cable length) of PD events as a function of the cable length.

In particular, after analysing the measuring data and the interpretation of information as combined with the test object data the following classification has to be provided in the test report: Cable system is PD-free; starting at higher than 1.7Uo electric stresses e.g. > 3Uo PD may start to ignite in accessories; in that case temporary over-stresses will not produce any PD inception in defects. Cable system contain discharging defects with PD inception voltage (PDIV) higher than 1Uo but lower than 1.7Uo; in that case due to temporary AC or switching over-voltages the PD development can be ignited and in the case of PD extinction voltage (PDEV) < Uo it can stay during network operation. Cable system contains discharging defect with PDIV< 1.0Uo; in that case in the cable insulation and/or cable accessories PD processes are active during network operation. Depending on the type of defect and the local electric field concentration besides knowledge about PD presence also the information about the decrease of PDIV, increase of PD magnitude are essential to evaluate the degradation process. Moreover, referring to asset management maintenance and replacement goal the test report should cover the following Information elements [8]: 1. 2. To describe the actual condition of a cable system. To provide information for further analysis of the cable status. 3. To give basic information for maintenance steps e.g [8]: - Condition of the cable system is OK, the next inspection can be performed within e.g. 5 till 10 years, - Condition of the cable system is doubtful; PD measurement should be performed within one year, - Cable system is not OK, immediately replacement is required. 7. CONCLUSIONS

magnitude cannot be predicted precisely, so it is necessary to perform calibration in several ranges. Performing one range calibration will affect the ability of the measurement to measure high PD magnitude or low PD magnitude. 3) Several points that should be taken in to account during performing measurement: a) In order to have a sensitive measurement, selection of proper PD range is important. b) To obtain sufficient information for PD analysis, PD measurement must be performed at several voltage levels. c) PD background noise should be measured to distinguish PD from external source and PD related defect in the cable system. 4) After performing PD measurement, relevant information about PD occurrence is collected and this data is called fingerprint of a cable system. The fingerprint is analysed to determine the condition of the cable system. 5) The final result of PD measurement is provided in the PD measurement report. The PD measurement report describes the condition assessment of the cable system and gives information of maintenance steps.

8.

REFERENCES

This paper discusses important aspects of PD diagnosis of MV power cables have been selected and discussed from the point of view of optimal insulation condition assessment. In particular the following can be concluded: 1) Test object definition is very important in interpretation of existence of PD occurrence in the cable system. Lack information about the component will result in ambiguity in interpretation of PD data. Relevant information about cable system, insulation types and accessories should be collected in order to obtain good interpretation of PD data. 2) The quality of PD calibration affects the whole quality of PD measurement. Due to the PD occurrence is a stochastic process, the PD

[1] F.J. Wester, Condition Assessment of Power Cables Using PD Diagnosis at Damped AC Voltages, ISBN 90-8559-019-1, Ph.D. thesis TU Delft, 2004. [2] Gulski, E, Lemke, E, Gamlin, M, Gockenbach, E, Hauschild, W, Pultrum, E . Experiences in partial discharge detection of distribution power cable systems. Cigre Electra, 2003, p 34-43. [3] 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 2002 Session, paper 104 [4] IEC 60270 Measurement of partial discharges, 2000 [5] IEC 60060-3 High Voltage test techniques Part 3: Definitions and requirements for on-site testing [6] F.H. Kreuger, Partial discharge detection in high voltage equipment, book 1969 Heywood, London, 1989, Butterworths, London [7] E Lemke et al. Practical aspects of the detection and localisation of partial discharges in power cables, Cigre brochure 297 [8] B. Quak, Information strategy for decision support in maintaining high voltage infrastructures, ISBN 978-90-8559-334-8, PhD thesis TU Delft, 2007 [9] Gulski, E, Smit, JJ, Quak, B, Wester, FJ, Groot, ERS . Decision support for life time management of HV infrastructures. In CEPSI'04; Proceedings of the 15th conference on electric power supply industry (pp. 1-8).