HVDC applications, electrical tree inception tests were performed before and ... For this purpose test objects of wire-plane electrode geometry were exposed to ...
2013 Annual Report Conference on Electrical Insulation and Dielectric Phenomena
Electrical Tree Formation as a Measure of Degradation Resistance in Polymeric Materials for HVDC Applications Le Wang1, Xiangrong Chen1, Libin Hu1,2, Stanislaw M. Gubanski1, Jörgen Blennow1 1
Dept. Material and Manufacturing Technology, Chalmers University of Technology, Hörsalsvägen 11, Gothenburg, 41296 Sweden 2 Dept. Electrical Engineering, Xi’an Jiaotong University Xianning West Road 28, Xi’an, 710049 China well as their comparison with results of ramped AC voltage tests, both performed on the same material are here reported.
Abstract - In the framework of elaborating a methodology for testing the resistance to electrical treeing of polyethylene (PE) for HVDC applications, electrical tree inception tests were performed before and after a thermal ageing on XLPE samples. For this purpose test objects of wire-plane electrode geometry were exposed to various voltage regimes. The obtained results show that a DC pre-stress followed by an application of voltage pulses of opposite polarity appeared most successful for a repeatable inception of electrical trees in the studied materials. Thin tree filaments were formed, distributing randomly at the vicinity of the wire electrode (i.e. high voltage electrode). More trees and longer branches were found in the samples having longer thermal ageing period. These results correlate well with results of in parallel performed tree inception tests under AC voltage ramping. It is therefore postulated that an AC electrical tree inception test can successfully be used for evaluating degradation resistance of materials for HVDC applications.
II. EXPERIMENT A.
Test Object Fig.1 shows the design principle and dimensions of the test object used in the performed experiments. A bent thin tungsten wire, 10 μm in diameter, is molded into the bulk of material sample before crosslinking. This wire acts as the high potential electrode, being externally connected via an attached copper tape. A bottom plane contact electrode provides the connection to the ground potential [7-9]. The electrode configuration used in these experiments differs largely from the traditionally applied needle-plane configuration [10, 11]. Its advantage is that the treeing process initiates at weak material spots along the wire surface, rather than being forced to start at the needle tip. Multiple trees are formed as well as a larger material volume is simultaneously exposed to the electric stress. It addition, the solution helps in avoiding some drawbacks of the traditional setup, namely by (i) reducing the mechanical strain caused during the needle insertion into material bulk and (ii) minimizing a risk for formation of gaseous cavities at the electrode-polymer interface. A possible formation of kinks at the wire electrodes during sample manufacturing is however a drawback that needs to be considered in result analyses if trees are formed at them [7-9].
I. INTRODUCTION The world’s increasing demand for electric power has promoted the tendency to use high voltage direct current based (HVDC) power transmission. However, many challenges within HVDC technology still await elucidation, such for example as evaluation of material resistance to degradation and breakdown caused by space charge build-up and electrical treeing. As the first paper on electrical treeing in solid dielectrics was published in the 50’s [1], the phenomenon has intensively been studied for more than fifty years. As usual with scientific work, researchers have been analyzing it from different points of view, raising various explanations, questions and conclusions [2-6]. Most of them however have focused on electrical treeing under exposure to AC voltage while only a limited number evaluate it under DC condition. In this paper, we concentrate on the possibilities to elaborate a methodology for testing the resistance to electrical tree inception in polymeric materials for HVDC applications. Several experimental procedures are introduced to test objects made of cross-linked polyethylene (XLPE) with wire-plane electrode geometry. These experimental attempts to repeatable inception of electrical trees have led to a conclusion that a test procedure based on a DC pre-stressing followed by an application of voltage pulses of opposite polarity appears to be a promising one. Test results obtained by using this method as
Fig. 1. Geometry of the XLPE sample; tungsten wire diameter - 10 μm, sample thickness - 3 mm
1 978-1-4799-2597-1/13/$31.00 ©2013 IEEE 510
B.
Experimental method for DC tree inception Up to now, four experimental methods have been established and reported in literature to incept electrical trees after exposure to DC stress, further called DC trees [12, 13]. These methods are summarized in the following: 1) Ramping DC voltage - Electrical trees can be incepted by applying a linearly increasing voltage to the sample [7]. As reported in [12], the tree inception voltage decreases with increasing voltage ramping rate, for both positive and negative voltage polarities. 2) Short-circuit - Pre-stressing the sample with a constant DC voltage for a known period of time, and then shortcircuiting it immediately after the voltage removal. The observed tree length can be taken as a measure of the travel distance of charges injected at the electrode-dielectric interface. 3) DC with superimposed impulse - Pre-stressing the sample with constant DC voltage for a fixed period of time and then superimposing an impulse on the DC voltage. The DC and impulse voltages may have either same or opposite polarities. 4) DC pre-stress plus impulse - Pre-stressing the sample with constant DC voltage (Vdc) for a fixed period of time (ta), turning off the voltage and leaving the sample open circuited for resting (tr), as defined in Fig. 2. Thereafter applying voltage pulses (amplitude Vi) having the same or opposite polarity as the pre-stressing DC voltage. There are four voltage combinations possible within this method. Among all these methods, the application of negative DC pre-stress in combination with positive impulses, shown in Fig. 2, appeared to be the most successful in yielding repeatable conditions for electrical tree inception in the tested material. The selected parameters of the tests were as follows: negative DC pre-stress voltage Vdc = 45 kV, pre-stress duration ta = 40 min, rest time tr = 2 min, amplitude of the positive impulse voltage Vi = 40 kV. Five pulses were applied after the DC pre-stressing, having front time of 1.5 μs and tail time of 70 μs. The time interval between adjacent pulses was 5 s. In the test, the negative DC and the following positive voltage pulses were applied at room temperature to ten in parallel connected samples (altogether immersed in a transformer oil bath).
C.
Experimental method for AC tree inception The AC tree inception test setup is sketched in Fig. 3. The system consists of a test cell having a transparent bottom, a microscope equipped with a CCD camera, a cold light source (to provide transmitted illumination of the sample) and a personal computer. During the test, the sample was immersed in transformer oil. A linearly ramped at a rate of 0.5 kV/s AC voltage was applied to the samples at room temperature.
Fig. 3 Schematic view of the experimental setup for AC tree inception tests.
Use of the microscope allowed recording real-time video sequences of the tree inception and growth. Normally several tree clusters incepted in each sample (at different times) along the tungsten wire electrode during each voltage ramp and the voltage level, at which the first tree appeared was recorded as the tree inception voltage. III. EXPERIMENTAL RESULTS A. DC tree inception DC tree inception tests were performed on five sample groups of the XLPE material and each of them consisted of ten test objects. Group 1, being the reference group, was neither degassed nor thermally treated after the manufacturing. Groups 2 to 5 were degassed, first for 96 h under vacuum and then for 72 h under atmospheric pressure. Groups 2 and 3 were degassed at 55°C, whereas Groups 4 and 5 were degassed at 80°C. After degassing, Groups 3 and 5 were thermally aged at 80°C for 200 h under atmospheric pressure. Microscope photos of the incepted DC electrical trees are shown in Fig. 4. The trees originated from the tungsten wire grow toward various directions (the ground electrode direction dominates). The short tree filaments have a single channel, in the length of some tens of micrometers. The longer ones are branched, have the length of several hundred micrometers and spread toward various directions. Some of them even grow perpendicularly to the externally applied electric field. Fig. 5(a) - (e) shows the bar charts of tree number (grey bars) in each of the tested samples, as well as the length (black bars) of the longest trees. The tree length was acquired by measuring the straight distance between the point of origin and the farthest point on the tree branch. As can be seen from Fig. 5, DC tree filaments were found in all the samples of Groups 3, 4 and 5. The largest number of trees appeared in one of the samples of Group 3, 19 trees were incepted randomly along the tungsten wire. However, 17 of them were single-branch short trees with their lengths ranging from 40 μm to 100 μm.
Fig. 2 Test procedure used for inception of DC electric trees.
2 511
A typical view of such tree filaments are shown in microscope photo of Fig. 4(a). For Groups 1 and 2, the trees appeared in seven of the samples of each of the groups. It is also important to stress that in Group 1, only one tree was incepted in each of the seven treeing samples, but one of them became as long as 1280 μm. The “#” mark in Fig. 5(a) - (b) indicates that the longest tree in the sample appeared at kinks on the wire electrode. A microscope photo exemplifying such a case is shown in Fig. 6. The effect is further discussed in section 4.
(a) DC trees growing toward ground electrode.
kink
Fig. 6 Tree incepted at a kink in the wire electrode.
B.
AC tree inception Six groups of XLPE samples were applied for estimation of tree inception level during tests with AC voltage ramping. Five of them were degassed and aged the same way as the groups used for DC tree inception tests. The other group was not degassed, but aged at 80°C for 200h under atmospheric pressure. Weibull plots and their parameters (i.e. scale parameter α and shape parameter β) of the AC tree inception voltages as well as the 95% confidence intervals are shown in Fig. 7 for each of the 6 groups. Group 1 shows the highest α value (α=39.0), which indicates the highest resistance to electrical tree inception, whereas both the degassing and the thermal aging of the remaining groups result in a strong reduction of the scale parameter α. Fig. 8 shows a microscope photo of AC electrical trees in a tested sample. AC trees grow randomly along the wire in clusters, propagating towards different directions, the ones incepted at the electrode loop growing faster than the others. Some of them can grow up to 600 μm within 10 s after the inception. However, DC and AC trees have significantly different appearance, while DC trees are of a clear branch type, AC trees exhibit a typical bush character.
(b) DC trees growing toward the direction opposite to the field.
Fig. 4 Appearance of DC trees.
(a) Group 1
(c) Group 3
(b) Group 2
(d) Group 4 (a) Group 1: without degassing and ageing (b) Group 2: degassed at 55°C (c) Group 3: degassed at 55°C, then thermally aged at 80°C for 200 h (d) Group 4: degassed at 80°C (e) Group 5: degassed at 80°C, then thermally aged at 80°C for 200 h
Fig. 7 Weibull plots and parameters of AC tree inception voltage (confidence intervals shown in the figure are 95%).
(e) Group 5 Fig. 5 DC tree number and maximum tree length in each of the investigated samples. Fig. 8 Appearance of AC tree in a ramp voltage test.
3 512
samples developed DC trees in Group 4 than in Group 2 (10 vs. 7). It is therefore likely that the degassing at higher temperature (80°C) made Group 4 samples more vulnerable to DC treeing. A similar conclusion can also be drawn from the AC results. Group 4 shows a smaller α value than Group 2 (22.8 vs. 25.8, see Fig. 7). By comparing graphs in Fig. 5(b) and (c), one may notice that Group 3 shows statistically more and longer trees. The 200 h long thermal ageing tends to weaken their ability to resist DC treeing. A similar trend can also be discovered in the results of AC tree inception test. Group 3 shows slightly lower α value than Group 2 (23.8 vs. 25.8, see Fig. 7). Also the difference between Groups 4 and 5 is not significant in both DC and AC results. Since the above presented analyses of the results of DC and AC tree inception tests show similar outcomes, it is postulated here that AC ramping test could be used as a robust screening method when ranking resistance to degradation in HVDC cable insulation materials.
IV. DISCUSSIOIN AND CONCLUSIONS A.
DC pre-stress plus impulse voltage method Though a clear explanation concerning inception mechanism of DC electrical tree is still missing, it is believed that formation of space charges plays here an important role. As a homo-charge is injected into the material from the electrode, to trigger electrical tree under DC voltage one has to overcome its effect on the locally acting electric field. All the DC tree inception methods described in Section 2 are based on this idea. The parameters of the test procedures, including those shown in Fig. 2, Vdc, ta, tr and Vi, can therefore influence obtained results. In particular, the level of negative pre-stress voltage Vdc controls electron injection, a larger Vdc value results in a larger amount of negative space charge in the electrode vicinity. In our experiments, DC trees could not be incepted in samples of Group 1 at Vdc lower than 40 kV. As ta is the time during which the injected electrons reach a stable state, it has to be in the order of several tens of minutes [10]. After the pre-stress voltage is turned off, the injected electrons dissipate, but it still takes them several tens of minutes. However, tr should be as short as possible. In addition, both the front time and the amplitude of the voltage pulse affect the DC tree inception. It is suggested that the front time should be shorter than 20 μs [12] and Vi large enough to incept the trees but not cause sample breakdown. In our case, if Vi was higher than 40kV, the probability of sample breakdown was largely increased. In addition, if only three voltage pulses were applied on samples of Group 1 after the pre-stressing, tree filament could not be found in any of them. Therefore, five pulses were applied in the procedure for obtaining comparable results.
REFERENCES [1] [2] [3] [4] [5] [6]
B.
Influence of the kinks at the wire electrode Kinks appear randomly along the wire in almost all the samples (see Fig. 6). The change of polymer morphology during the sample manufacturing and annealing is likely responsible for their formation [7-9]. Since the electric field at the tip of a kink may be up to 70% higher as compared to the smooth part of the wire electrode, DC and AC electrical trees often tend to incept first at this point. In all the samples of Group 1 the longest trees incepted at the kinks, as exemplified in Fig. 5. At the same time, only single samples in Groups 2 and 3 as well as 5 samples in each of Groups 4 and 5 had the longest tree at the kink. For excluding the influence of trees incepted at the kinks, these should not be considered in performed analyses. The tree inception is not only dependent on the electric field at the electrode surface but also on the material morphology in its vicinity, since both the degassing and the thermal treatment can influence it. The observed change in the number and the length of DC trees can therefore explain this way.
[7] [8]
[9]
[10] [11] [12] [13]
J. H. Mason. “The deterioration and breakdown of dielectric resulting from internal discharges”. Proceedings of IEE, 1951, 98 (1): 44-59. Densley R J. “An investigation into the growth of electrical trees in XLPE cable insulation”. IEEE Trans. on Electr. Insul., 1979, EI14(3):148-158. K. H. Stark, C .G. Garton. “Electrical strength of irradiated polythene”. Nature, 1955, 176 (4495): 1225-1226. M. Ieda and M. Nawata “A consideration of treeing in polymers”. 1972 Annual Report CEIDP, Washington D.C,1972:143-150. T. Tanaka “Charge transfer and tree initiation in polythene subjected to ac-voltage”. IEEE Trans. on Electr. Insul., 1992, 27(3): 424-431. K. Wu, L. A. Dissado, “Model for electrical tree ion in epoxy resin”, IEEE Trans. on Dielectr. and Electr. Insul., Vol. 12, No. 4, August, pp.655-668, 2005. Anette Johansson, “Detection of electrical treeing in XLPE exposed to AC and DC stress”, Master thesis of Electric Power Engineering, Chalmers university of Technology. A. B. Johansson, S. M. Gubanski, J. H. M. Blennow, M. Jarvid, M. R. Andersson, B. Sonerud, V. Englund and A. Farkas, “A versatile system for electrical treeing tests under AC and DC stress using wire electrodes” presented at Jicable 2011, Versailles, France, 2011. M. Jarvid, A. B. Johansson, J. H.M. Blennow, M. R. Andersson and S. M. Gubanski, “Evaluation of the performance of several object types for electrical treeing experiments”, submitted to IEEE Trans. on Dielectr. and Electr. Insul. A. C. Ashcraft, R. M. Eichhorn and R. G. Shaw, “Laboratory studies of treeing in solid dielectrics and voltage stabilization of polyethylene”, IEEE ISEI, Montreal, Que., Canada, pp. 213-218, 1976. G. Jiang, J. Kuang and S. Boggs, “Critical parameters for electrical tree formation in XLPE”, IEEE Trans. on Power Deliv., vol.13, pp. 292-296, 1998. M. Ieda and M. Nawata, “DC treeing breakdown associated with space charge formation in polyethylene”, IEEE Trans. on Electr. Insul., Vol. EI-12, No.1, pp.19-25, February, 1977. F. Noto, N. Yoshimura, T. Ohta, “Tree initiation in polyethylene by application of DC and impulse voltage”, IEEE Trans. on Electr. Insul., Vol. EI-12, No.1, pp. 26-30, February, 1977.
C. Similarity between DC and AC results Although the morphologies of trees formed under DC and AC conditions are different, some similarities of a general type can be found between the results of both tests. For example, Group 1 samples show stronger resistance to electrical tree inception under both DC and AC voltages. More
4 513
