Key Words: polymer insulation, interface, testing, monitoring, partial discharge, termination, cable, silicone rubber, high voltage, XLPE olymer technology has ...
*
F
E
A
T
U
R
E
A
R
T
I
C
L
E
Dealing with Interface Problems in Polymer Cable Terminations Key Words: polymer insulation, interface, testing, monitoring, partial discharge, termination, cable, silicone rubber, high voltage, XLPE olymer technology has shown good performance and has certain advantages over ceramic technology in irisulators. However, it often comes with a potentially weak point, i.e., the interface between two solid insulators. In The Netherlands, a cascade of 150 kV termination breakdowns occurred several years ago. The present paper gives the results of the research and precautions that were undertaken after those events. The problems started at the interfaces of these components, as will be explained. The investigation and monitoring of terminations is also described. The information in this paper on interface testing cells was recently presented at the ISEIM '98 (International Symposium on Electrical Insulating Materials) [11.
Interfaces As a consequence of the wider introduction of polymer technology in cable systems, combinations of solid insulating materials in components are common. In most tradlitional systems, insulators consisting of impregnated paper and oil or fat face each other in joints and terminations. [n polymer technology, interfaces can occur between insulating polymers such as crosslinked polyethylene (XIJE), silicone rubber (SIR), ethylene-propylene rubber (EPK) and epoxy resins. Such interfaces exist between cable insulation and rubber stress cones in joints and terminations or between rubber stress cones and epoxy parts in joints. The interfaces in such components frequently run partly parallel to the electrical field. As a rule of thumb, the inttrface of two insulators is usually weaker than the bulk of each individual insulator. Moreover, in the case of solid insulators, there is the risk of gaps. The gas or vacuum in such a gap in combination with the surrounding insulating materials generally provides a locally lower insulating strength. Achieving reliability in interfaces that run (partly) parallel to the electric field thus requires great care. Where possible, situations of high tangential electric stresses (i.e., parallel to interfaces) should be avoided. July/AuguSt 1999 -Vol.
15, NO.4
Robert Ross EM, Transmission 6 Distribution Power
A cascade of breakdowns in a series of t.50 kV terminutions on one duy in 1993 caused a major blackout in The Netherlrrlnds. Experience with Interface lproblems The issue of interface reliability is not an academic exercise. A cascade of breakdowns in a series of 150 kV terminations on one day in 1993 caused a major blackout in The Netherlands. The blackout even became the subject of a sociological study on the impact of such events on modern society. In other countries interface problems have also occurred in cable joints. Fig. 1 shows the interface of the XLPE insulation and the SIR stress cone in a 150 kV cable termination. Both parts bear the marks of the electrical trecing that took place during service. The two imprints are each other's negatives. A breakdown ultimately followed the discharging at the interface. Similar treeing patterns were found at the interfaces of terminations that were taken out of service before a breakdown could occur. This shows that such electrical treeing takes place in a matter of days or months rather than minutes. It is noteworthy that the treeing started at the interface without any direct connection to any of the electrodes. Electric field calculations confirmed that the discharges occurred where the electric field component along the interface was highest. Furthermore, testing in a climate chamber showed that terminations that have already discharged show a large increase in discharge activity during temperature change. This enhanced discharging is associated with XLPE
0883-7554199/$IO.0001999lEEE
5
ease of handling, etc. Recommendations to improve the system have been discussed but are beyond the scope of the present paper. In order to prevent outages due to interface problems, two main initiatives have been developed: investigations to understand the cause of interface problems and the introduction of a dedicated partial discharge monitoring method. These subjects are addressed in the following paragraph s.
Interface Investigations Fig. 1 Discharge pattern at the interface o f a failed 150 kVcable termination. The SiR stress cone has been cut, At the bottom of the figure, one half is shown w i t h electrical treeing and breakdown marks. T o p half of the figure shows the XLPE insulation with the breakdown channel. See also Fig. 2. Area of HighestTangential Field and of ElectricalTreeing
SIR Stress Cone
XLPE Cable
Fig. 2 Stress cone and position of highest electric field Front Transviews
Side Transviews
The investigations into the interface problems comprised studies on the design of the terminations, electric field calculations and research to determine factors with a critical influence on reliability.
Design and Field Catcadations The design of the terminations is such that the highest tangential field is in the range of < I kV/mm (typically 0.4-0.7 kVimm) depending on surface smoothness. The normal position of the highest tangential field at the interface of the XLPE cable insulation and the SIR stress cone is about where the deflector top would be projected on the XLPE surface (see Fig. 2). The electric field depends on the design, concentricity of the cable, mounting, and both the cleanliness and contact of surfaces. With imperfections at the interface, the tangential electric field may run up 1-5 kV/mm, which then equals about the intrinsic strength of the interface. Obviously, the interface needs to be handled with care and adverse influences must be suppressed adequately.
Research and Testing of Interface Reliability
Fig. 3 Interface testing cell for multi-stress ageing. Electrode configurations: see text. Left-right motion of a block in front views preserves the field at hot spots.
cable and SIR parts shifting along each other due to the difference in thermal expansion coefficients. During mounting a grease is used, which is largely absorbed by the SIR later. The interface is considered to be rather dry. The discussions following the failures addressed design, electric fields, mounting quality, contamination,
6
Since the failures in 1993, research and test development have been undertaken in The Netherlands. in order to study the strength of interfaces, experiments were carried out on terminations for various voltage levels. However, the many factors that could have an influence on interface reliability would mean that an unacceptably large number of terminations would be involved. Moreover, the feeling grew stronger that the phenomenon would particularly occur in large terminations. Therefore, practical and economical arguments pointed to the development of interface sample cells that would allow quick results with a wide variety of experimental parameters. Worldwide, various research groups have also started to develop interface testing cells [2].Within CIGRE WG 15-10 a Japanese group investigated the merits of a number of test cell designs that ranged from easy-to-produce models to miniature joints. In 1995 the discussion was broadened to an international level involving CRIEPi, Toyohashi UT, Ontario Hydro and KEMA. in 1996 universities from Germany and Denmark joined the discussion on test cell requirements. Meanwhile the subject draws more and more attention. The group of involved parties is growing. In Japan and The Netherlands national groups of utilities, industry, test laboratories and universities have been established to discuss interface problems and give input to CiGRE. The purpose of IEEE Electrical Insulation Magazine
such test cells is to study or test the design, the capability of withstanding external influences, tolerances in mounting, acceptable levels for defects, the material’s (in)compatibility, the influence of external parameters like pressure and temperature, the effect of grease or pollution, etc.
Test Cell Requirements and Design In 1996 a list of test cell requirements was established by CIGRE WG 15-10 (cf. [3]). Interface testing cells should: 1-have a simple configuration that is easy to reproduce; 2- have no metal electrode surfacing at the interface;
5 - tip shape and overlap of electrodes (Figs. 3b and 3c). The tangential electric field distribution is shown in Fig. 4 by means of the equipotential lines in the interface. These concern the crossed rods and finger-rod geometries (Fig. 3). In the calculations, the lower block always consisted of XLPE ( ~ ~ 2 .with 2 ) an earth electrode and the upper of SIR ( ~ ~ 3 . 1with 5 ) a high voltage electrode. The interface area is 50 x 50 mm2. The earth electrode runs vertically in the plots. Electric fields along practical interfaces are in the 0.5-2 kV/mm range. Cell geometry and test voltage should allow such fields without partial discharge (PD) levels being equaled by corona or by treeing a t other positions. Other-
3- allow various defects to be introduced; 4- enable one to study mechanical pressure effects; 5- enable one to study surface roughness effects; 6- enable one to study the effect of silicone oil or other liquid insulants; In the course of research at KEMA, it was considered necessary to add the following item: 7- enable one to study shear effects (including motion and rubbing). The last item was added because of the increased discharge activity in terminations under temperature change and the accompanying me4a: Crossed rods; 4b: Zoorning in on top right hot spot of 3a chanical effects. Of particular importance is the Size: 10mrn x 10mm.Voltage steps are .03xVt hl=hp = 1 mm, dl=dp = 4 mm possibility of having motion parallel to the highest tangential electric field. After two earlier generations of sample cells [4], a third generation was designed [l]. The main idea of the design is to obtain a constant electric field, particularly at the hot spot(s) of the interface when two materials rub along each other. Here, hot spots relate to the highest tangential field, i.e., along the interface. The position of the hot spots is defined relative to the top electrode (Fig. 3). Preferably, shear in the direction of the tan4d: Zooming in on top right hot spot of 3c hl=h2 = 5 mm; dl=d2 = 4 mm Size: 10mrn x 10mm.Voltage steps: OlxV, gential electric field at the hot spots of the interface must be possible. Fig. 3 shows three types of the cell. In each case, the cell consists of twoblocks of insulating material. In at least one, an electrode runs parallel to the interface to provide invariance of the electric field with respect to motion. The second electrode is either a crossing rod also parallel to the interface (Fig.3a), or a finger electrode (Fig. 3b), or both electrodes run parallel until a little over half the samples (Fig. 3c). 4e: Finger rod; The samples are easy to produce in large quanhl=h2 = 1 rnm; d l 4 2 = 4 mm tities by casting an insulating material around a metal rod or tube. The parameters that control the Fig. 4 Equipotential lines at interfaces of variousgeometries. (V, = test voltage). 4a: crossed rods; 17, = h , = 1 mm; d , = d, = 4 mm; 4b: zooming in on top right hot spot electric field are: of Fig. 4a; size: 10mm x IOmm. Voltage steps are 0.03 x V,; 4c: crossed rods; h , = h, 1-voltage (V); = 5 mm; d , = d, = 4 mm; 4d: zooming i n on top right‘ bot spot ofFig. 4c; size: 1 Onlm 2- dielectric properties of the materials; x 10mm. Voltage steps are 0.01 x Vi;4e: finger-rod; h , = h , = 1 mm; d, = d, = 4 mm; 3- electrode-interface distances (h, and h2); 4f: zooming in on top hot spot of Fig. 4e; size: 10mm x 10mm. Voltage steps ure 0.016 x v, 4- diameters of the electrodes (d, and d2); JUly/AUgUSt I999 -Vol.
15,NO.4
7
wise, techniques have to be used to distinguish one source from another. Preferably the test voltage should be in the medium voltage range.
Crossed Rods Figs. 4a and 4c show the equipotential lines at interfaces with h,=h,=l mm and h,=h,=5 mm. Figs. 4b and 4d zoom in on the top right hot spot of the respective cases. Here, the electrode diameters are all the same: d,=d,=4 mm. The four hot spots are found along the bisectors (line under 45" from the center) between the rods. Short distances h, and h, produce hot spots close to the center. With a larger h, and h,, the equipotential lines run more parallel to the high voltage electrode. The larger h, and h,, the more the surroundings of the test cell can influeiice the field distribution at the interface of course. A tangential field of 1 kV/mm at the hot spots requires a test voltage of 9.3 kV and 48 kV for the respective cases. The maximum field at the electrode becomes 6.6 resp. 10.7 kV/mm for XLPE and 4.6 resp. 6.4 kV/mm for SIR. Motion and tangential field are parallel for two hot spots with shifts under about 45" to an electrode axis and about perpendicular at the other hot spots. The latter is unlike the situation in 1.50kV terminations that were suffering from interface degradation. There the shifts and the tangential fields were parallel. With a larger h, and h,, motion and field become reasonably parallel with shifts along the earth electrode (Figs. 4a and 4c). However, a much higher voltage is necessary to obtain significant tangential electric field strengths. The crossed rods model can be used for motion experiments, but it seems best in static experiments. The cell is fairly simple to prepare and gives four hot spots. The latter feature can have the advantage of better statistical significance.
Finger-Rod Fig. 4e shows the equipotential lines in the interface for a finger-rod geometry, while Fig. 4f zooms in on the top hot spot. Again, d,=d,=4 mm and h,= h,= 1mm. This geometry and that of Fig. 3c give parallel motion and tangential field at hot spots with shifts along the earth electrode. A tangential field of 1kV/mm at the hot spots requires avoltage of 12 kV. The field strengths at the electrodes are 8.9 kV/mm for XLPE and 6.9 kV/mm for SIR. The finger-rod geometry seems convenient for ageing involving shear effects. Positioning of the finger requires more care than for the crossed rods configuration. However, this geometry seems quite convenient for shear effects. Therefore, particularly this finger-rod geometry is very promising for motion along the highest tangential electric field. Currently, the most convenient shape of the tip of the finger electrode is being studied.
Monitoring Interface Reliability As outlined above, the breakdown of the terminations is preceded by discharge activity in the interface. This also gives the possibility of pre-breakdown warnings. In order to enable timely replacement and prevent outages, a PD (partial dis8
:i
50 Hz Current Lead
Cable Lead Sheath
Fig. 5 Set-up measurement on a termination, from [6/
charge) monitoring program for 150 kV terminations was set up. A new, patented technique was developed for the diagnosis of joints and terminations, dubbed HFPD (High Frequency PD) or VHF PD (Very High Frequency PD) detection [S,61. Techniques such as conventional and acoustic PD measurements have several disadvantages. The former operates in the band of < 1MHz and suffers from a significant noise level generally. The latter becomes of limited use, when PD sites are acoustically hidden or give low PD levels. Such levels may already be an indication for serious degradation.
HFPD Measurements The HFPD technique differs from conventional PD measurements by processing signals in the range of 100-500 MHz range. The frequency range makes the technique useful for short cable pieces and for accessories (terminations or joints) like the 150 kV terminations. In longer lengths of cable, high frequency signals become attenuated and will not be measurable. While being a disadvantage for PD analysis for full cable lengths, the rapid reduction of HFPD signals is a major advantage with accessory testing. The reason is that HFPD activity from other accessories in the cable hardly contributes to HFPD signals measured at the termination. Thus, if HFPD activity is measured close to a termination, then either the termination or the first part of the cable discharges. The termination is usually the most likely source, because XLPE cable fails fast after PD becomes measurable, IEEE Electrical Insulation Magazine
while terminations have shown PD activity over a period of weeks or months before breakdown. The HFPD technique picks up pulses in the frequency range of a radio signal at interruptions or openings of the earth screen (Fig. 5 ) . A coax cable is connected to a sufficiently large (local) gap in the earth screen and therefore no connection with high voltage parts is needed. The applied terminations have an interruption in the earth screen over which the HFPD pulses are measured. The possibility of picking up signals from the earth screen is a great advantagse, which makes it fairly easy to monitor (HF)PD activity in terminations on-fine. Coax cables and capacitive voltage divitlers are installed when the cable is de-energized.
Gaining Practical Experience About 100 terminations have been monitored with the new technique. Normally, the noise level is as low as 30 to 4t0 pC in the high frequency range of 100-500 MHz. With the classic method, the noise level is 100-4000 pC. Normally, the local situation allows one to measure six terminations per day. One termination failed after two years, showing a large increase in the HFPD level with its annual check, which was months prior to the breakdown. With a failure analysis, the interface showed an electrical treeing pattern at the interface as discussed above. The HFPD technique is considered mature and able to detect PD activity in a convenient way. The method is also employed in a pre-qualification test of a 400 kV cable circuit at KEMA. Here sensors are also positioned at the joints. Interestingly, the previously described interface sample cell was used to calibrate the measuring system. It was also suggested to include PD measurements li’ke this in after-laying ac tests for HV (High Volta.ge) and EHlV (Extra HV) cable systems [7].
Conclusions Interfaces with large tangential electrical field components (i.e., parallel to the interface) require great care. Particularly with thermomechanical action, the discharges may be considerable once PD activity has started. In order to study factors that influence the interface reliability, an interface sample cell has been designed. A special feature is the ability to determine the effect of surfaces shifting along each other. Apart from
July/AugUst 1999 -Vol.
15,NO.4
laboratory studies, interface reliability can be monitored on a routine base by a dedicated technique, dubbed HFPD or VHF PD measurements. The method gives a better signal-to-noise ratio than classicalPD measurements and is particularly sensitive to the accessory at the measuring site.
Robert Ross graduated froin Utrecht: University (UU) in physics and mathematics. He earned his Ph.13. in physics and chemistry on water treeing in XLPE. From 1983 until 1986 he worked as an educational assistant in signal processing at UU. Since 3986 he has worked at KEMA holding various positions in research and consultancy. From December 1992 hie was on a one-year sabbatical leave at the National Institute of Materials and Chemical Reseach in Tsukuba, Japan, workiing on CGO-molecules. In 1994, he co-founded W O (nonprofit Institute for Science and Development) and is its president. Here, work is carried out on Weibull statistics. At KEMA he currently is a senior consultant and product manager for materials research. He is member of various groups in CIGRE, IEEE, IEC, and IEE.
References 1. R. Ross and M.G.M. Megens, “An Interface Testing Cell for Multi-Stress Ageing,” Proc. ISEIM’98, Toyohashi, S8ept.27-30, 1998, paper P2-34. 2. T. Tanaka, “Polymer Interfaces Associated with Electrical Insulation Systems,” presented at the CIGRE SC 15 colloquium, Bedford, Mass., USA, Aug. 18, 1997. 3. M. Nagao, N.Akagi, M. Kosaki, and T. Tanaka, “Model Specimens for Testing Interfacial Electrical Insulating Properties in EHV-Extruded Cable Splices and Preliminary Kesults-Specimen with Sphere Electrode (EPR-epoxy and XLPE interface),” 1997, CIGRE WG15-10 doc. 1.5-10-Nagao-01-97. 4. R. Ross, “Interface Modeling,” KEMA memo, 1997, CIGRE WG15-10 doc. 15-10-Ross-01-97, 5. P.A.A.F. Wouters, P.C.T. van der Laan, E. Hetzel, and E.F. Steennis, “New On-Line Partial Lhcharge Measurement Technique for Polymer lnsulated Cable and Accessories,” Proceedings 8th ISH (lnt. Symp. on High Voltage Eng.), Yokohama, Aug. 23-27, 1993. 6. E. l’ultrum, E.F. Steennis, and M.J.M. van Riet, “Test After I.aying, Diagnostic Testing Using Partial Discharge Testing at Site,” Proc. ClGRE 1996 joint session, 15/21/33, paper 12. 7. N. van Schaik, H.N.E. Sibbel, and O.J. Groen, “011-Site Test after Laying with AC Voltage of Significant Lengths of HV and EHV XLPE Cable Systems,” Proceedings CEPSI ’98, Thailand, 1998.
9
