Investigation of the Residual-charge Technique for Diagnosis of Water-tree Deteriorated Cross-linked Polyethylene Cable Dr. Uchida Katsumi*, Youichi Katou*, Dr. Hiroyuki Kon, Dr. Kazuo Watanabe and Takenori Tokutake Water-tree deterioration in XLPE cable is one of the factors that reduce breakdown strength. To assure stable power supply, an effective technique has been needed to estimate the degree of the water-tree deterioration in XLPE cable. We have been investigating the application of the residual-charge technique for this purpose. We have confirmed the effectiveness of the residual-charge technique on the diagnosis for water-tree deteriorated XLPE cables, and possibility of application of oscillating wave (OSW) to the residual-charge technique, which is effective on the field measuring because it leads to downsizing of the apparatus. We also propose a new technique to estimate the degree of watertree deterioration. This technique could make it possible to detect the existence of the localized and heavy water tree deterioration in XLPE cable. 1. Introduction It is well known that there is a possibility of watertrees formation in cable insulation, if cross-linked polyethylene (XLPE) cable is used under the wet conditions for a long time. Water trees influence the dielectric performance in XLPE cable such as breakdown strength. It has been about 3 years, which is thought as a life span of XLPE cable, since XLPE cable without water impervious layer started to be used 1). The possibilities of water-tree formation and degradation of breakdown voltage due to its formation are considered in case of old fashioned and poor manufacturingprocess-cables subjected to a long-term immersion in water and service aging. The development of the method to estimate the degree of water-tree deterioration in XLPE cable has been highly necessitated. There are currently several methods to estimate the degree of deterioration due to water trees. It has been reported that conventional methods such as leakage current,tan δ etc. are difficult to apply to 22 to 33 kv XLPE cable and that the estimated value has little correlation with breakdown voltage. On the other hand, new methods have been proposed to solve the problem. We have been investigating the residual-charge technique as a method for diagnosing for water-tree deterioration of XLPE cable. The effectiveness of the residual-charge technique and new estimation techniques are reported. * Chubu Electric Power Co., Inc. 2. Principle of Residual-charge Technique Fig.1 and 2 show the principle of this method and the pattern of voltage application, respectively 2). DC voltage is applied to a cable as a pre-stress. Charges DC high voltage generator 6.6kV : 15kV 22 kv : 3kV R AC testing transformer Low-pass filter Cable Fig. 1. Principle of Residual-charge Technique. 5min DC Cd Rd ~ 1kV/s 1min 4s AC Output ~ 1kV/s Fig. 2. Pattern of Applied Voltage in Residual-charge Measurement. Fujikura Technical Review, 21 65
are accumulated in water trees by DC voltage application. AC voltage is applied to it after short-circuiting. Charges accumulated in water trees are released during AC voltage application, which produce a DC current component. Though the current flowing during AC voltage application includes both AC and DC current components, DC component can be detected by using a low-pass filter. Fig.3 shows a typical result. Such a current signal shown in Fig.3 can be detected if the XLPE cable contains water trees. The residual charge is obtained by integration of a current signal. 3. Experimental 3.1 Samples Test cables were prepared by frequency-accelerated aging with hot-water immersion by using XLPE insulated cable of 6.6 kv with conductor of 6 mm 2 and 22 kv with conductor of 2 mm 2. The details of the accelerated aging conditions are listed in Table 1. For convenience, we classified the samples as Type 6-1 ~ 6-4 and Type 22- and 22-1. The water-tree distributions of the samples aged as above are shown in Fig.4. All of the water trees observed in the samples were bow-tie type. It can be said that the degree of deterioration of is higher than that of, because the total number of water trees increases as aging time. Classification Table 1. Conditions of Frequency Acceleration Condition of applied voltage Condition of temperature (Equivalent time to 6 Hz) 6.6kV XLPE cable / 6mm 2, t = 4 mm, length of the cable : 1 m, length of immersed region in hot water : 4. m Type 6- Frequency : 11,98 1-3kHz Applied voltage : 8 C 36,53 1kV 55,533 117,583 22kV XLPE cable / 2mm 2, t = 7 mm, length of the cable : 2 m, length of immersed region in hot water : 4. m 1 1 Type 22- Frequency : 1-3kHz 8 C Applied voltage : 2kV 37,3 Current (na) 5 5 Charge (nc) Table 2. Conditions of Residual-charge Measurement Classifications DC voltage AC voltage Pattern of AC voltage application 12 Time (s) Fig. 3. Example of Residual-charge Measurement. 6.6kV XLPE cable 22kV XLPE cable 15kV/5min 3kV/5min 3.8kV 12.7kV Increasing rate of AC voltage : 1kV/s Keeping time of AC voltage : 6s. Decreasing rate of AC voltage : 1kV/s 5, Number of water trees 4, 35, 3, 25, 2, 15, 1, 5, 21~5 11~2 31~4 Number of water trees 4, 3, 2, 1, 21~5 11~2 31~4 Type 22- Classification of water tree length ( m) (a) 6.6kV CV Classification of water tree length ( m) (b) 22kV CV Fig. 4. Distribution of Water Trees in Frequency-accelarated Aged Cable. 66
3.2 AC Breakdown Test Conditions AC breakdown tests were carried out at about 8 C. 3.3 Residual-charge Measurement Residual charge was measured basically under the conditions listed in Table 2. 4. Results 4.1 AC Breakdown Tests The results of AC breakdown tests are shown in Fig.5. Breakdown voltage decreased with aging time. Aging time means the equivalent time to service frequency. As mentioned above, cables were deteriorated by water trees as aging time, therefore, the decrease in breakdown voltage is thought to correspond to water-tree deterioration. 4.2 Residual-charge Measurement The aging time dependence of residual charge is shown in Fig.6. Residual charge was first observed in, and increased with aging time. This tendency was also observed in the Type 22 samples. AC breakdown field strength (kv/mm) Residual charge (nc) 5 4 3 2 1 Type 6- Type 22-2, AC testing was carried out at 8 C 6.6kV cable 22kV cable 4, 6, 8, 1, 12, 14, Fig. 5. Agingt Time vs. AC Breakdown Field Strength. 5 4 3 2 1 Type 6- Type 22-6.6kV : DC 15kV/5min, AC3.8kV 22kV : DC 3kV/5min, AC12.7kV 2, 4, 6, 8, 1, 12, 14, Fig. 6. Aging Time vs. Residual Charge. Considering the time dependence of the distribution of water trees in these samples, the tendency of residual charge with aging time corresponds to the deterioration due to water trees. 4.3 Correlation of Residual Charge and Breakdown Strength Fig.7 shows the correlation between residual charge and breakdown field strength of the 6.6 kv and 22 kv XLPE cable. Residual charges shown in Fig.7 are measured with AC field strength of 1. kv/mm. Despite the different rated voltage of cable, 6.6 kv and 22 kv, residual charge has a good correlation with breakdown field strength. This indicates that the residual-charge technique is an effective method for diagnosing water-tree deterioration in XLPE cable. 5. Residual-charge Technique in Field Diagnosis 5.1 Application of Oscillating Wave (OSW) to Residual-charge Measurement It is necessary to prepare the AC high-voltage source with quite large electrical capacity, which becomes physically large, to measure residual charge of the cables spanning a few km. The size of AC high-voltage source is one of the factors that make the residual-charge measurement in field difficult. We investigated the possibility of substituting an oscillating wave (OSW) for AC voltage source. It is expected to be effective on field measurement because electrical capacity of OSW generator is smaller which leads to more compact equipment than the AC high-voltage generator of service frequency. Applying OSW to residual-charge technique, some drawbacks have to be resolved, such as the effect of OSW on water trees. It has been reported that OSW application to water-tree deteriorated XLPE cable has a possibility to initiate an electrical tree from a water tree 3). AC breakdown field strength (kv/mm) 5 4 3 2 1 1 2 3 Residual charge (nc) 6.6kV : 6mm 2 22kV : 2mm 2 Fig. 7. Residual Charge vs. Breakdown Field Strength. 4 5 Fujikura Technical Review, 21 67
1 water electrodes 2mm 2mm 1 water electrodes Outer semiconducting layer Thickness of insulation : 3mm 2mm 1mm XLPE Inner semiconducting layer Conductor Fig. 8. Location of Water Tree Modeled Electrodes. 3 Table 3. Results and Conditions of Investigation on OSW Voltage (kv) 2 1 1 2 Classification Tno.1 Tno.2 Sample Depth Number of locations Test conditions Occurrence of electrical tree from water electrode 1mm 1 1 kv (peak voltage) 1mm No 2mm 1 1, times 2mm No 1mm 1 1 kv (peak voltage) 1mm No 8 times increasing at a 2mm 1 rate of 1kV/1 times 2mm Yes 3 2 4 6 8 1 12 14 16 18 2 Time (ms) Fig. 9. Applied OSW Waveform. 5.1.1 Effect of OSW on Water Trees (1) Samples We prepared the sample cables with water electrodes modeled on a water tree generated from the outer semiconducting layer 3). The procedures are as follows, q Treeing needle is pricked from the outer semiconducting layer into cable insulation to the depth of about 1mm and 2mm. w Treeing needle is drawn out in water. e AC voltage is applied to the cable given the above treatment under the condition of 1kVrms for 24 hrs. in water of 8 C. The details of the prepared sample are shown in Fig.8. (2) Experimental The OSW shown in Fig.9 was applied repeatedly to the sample prepared as described above to investigate the effect of the OSW on the water trees in cable insulation. The effect on the water trees is determined by monitoring the sliced samples for occurrences of electrical trees initiated from water trees. (3) Results The results are listed in Table 3. No breakdown and electrical trees from water electrodes were observed after 1, times applications of OSW with a peak voltage of 1 kv. This indicated that OSW application with a peak of 1kV has no effect on water trees with a length below 2. mm from the outer Fig. 1. Electrical Tree from Tip of Water Electrode. semiconducting layer in the 6.6 kv XLPE cable. As for water trees from the inner semiconducting layer, which were thought to be the most harmful, it is inferred from the calculation of field strength that OSW application with a peak of 1kV has no effect on the water trees with a length of 1. mm. After 8 times OSW application with a peak of 1kV, increasing up to a voltage at a rate of 1kV/1 times leads to a breakdown at a peak voltage of 4 kv. Electrical trees were observed at 2mm-depth water electrodes (Fig.1), but not at 1mm-depth water electrodes. This result supported the above indication. 5.1.2 Residual-charge Measurement with OSW Next, we applied an OSW to residual-charge measurement. The pattern of voltage application was basi- 68
Residual charge (nc) 1 5 Type 6- Type 22-2, 6.6kV : DC 15kV/5min., OSW of 1kVpeak 22kV : DC 3kV/5min., OSW of 2kVpeak 4, 6, 8, 1, 12, 14, Normalized current ( / max value) 1.5 1.5 2 6s 4 6 Time (s) 1kV Pattern of AC voltage application 8 1 12 Fig. 11. Results of Residual Charge by Using OSW. Fig. 12. Variation of Residual-charge Signal. cally the same as shown in Fig.2, the difference was only to use an OSW as a substitution of AC voltage. Applied OSW had a peak voltage of 1kV with a frequency of 5-6Hz for the 6.6 kv XLPE cables and a peak voltage of 2 kv with the same frequency as the above for the 22 kv XLPE cables. The peak voltage for 22 kv XLPE cable was determined by comparison of the differing thickness of the 6.6kV and 22kV XLPE cables; that is, the thickness of 22kV XLPE cable is nearly twice thick as the 6.6 kv XLPE cable. The OSW with a peak voltage of 1 kv is thought to have no effect on water trees in these samples, because samples were Type6- ~ 6-4 and Type22- and 22-1. The longest water tree is about 5µm in these samples. The results are shown in Fig.11. Residual charge could be detected by using OSW, and the aging-time dependence of residual charge (Fig.11) showed the same tendency as in Fig.6, but the amount of residual charges was about.3 times less than that of the results by using AC voltage application of 1. kv/mm. These results indicate the effectiveness of substituting an OSW for AC voltage in residual-charge measurement. 5.2 Method for Long Cable Span The above results were obtained by testing the cable with a length of 1-2 m, and the deteriorated region due to water trees was about 4 m. This length (4 m) corresponded to the length immersed in hot water during frequency acceleration aging, and the distribution of water trees was almost uniform within this region. But considering the cables in field, the distribution of water trees is not thought to be so uniform. Generally speaking, it is difficult to discriminate the localized and heavy water-tree deterioration and light and uniform water-tree deterioration by using the conventional residual-charge measurement, since the obtained residual charge depends on cable length. In the case that the residual-charge measurement is carried out by using the cables of the same length with localized and heavy water-tree deterioration or light and uniform water-tree deterioration, it could lead to a conclusion that these two cables have almost the same degree of water-tree deterioration, even though the breakdown strength of the latter is lower than that of the former. We have been developing the new technique to discriminate them. 5.2.1 Relationship between Water-tree Distribution and Residual-charge Signal Fig.12 shows the current signals of residual charge from samples of ~ 6-4 obtained by using AC voltage application as increasing up to 1 kv at a rate of 1kV/6 sec. The waveforms in Fig.12 are normalized by the maximum value of each signal for the purpose of waveform characteristics comparison. The signal of increased as AC voltage and then decreases monotonically. On the other hand, the signal of and 6-4 did not decrease monotonically after increasing voltage application; that is, both signal had a moderate peak at higher AC voltage. This indicates that there is a component of residual charge that requires application of a higher AC voltage to be released from water trees in heavily deteriorated cables such as and 6-4. As shown in Fig.4, has a large amount of water trees of about 5 µm -length, but few of length exceeding 5 µm, and the number longer than 1 µm drastically increases in and 6-4 in comparison with. Considering the difference in the water-tree distribution between and and 6-4, the component of residual charge which requires a higher AC voltage to be released is thought to correspond to long water trees. How high a voltage is required for a given length is not clear at this stage, and more investigation is needed. Fujikura Technical Review, 21 69
5.2.2 New Method for Detecting Long Water Trees As mentioned above, residual charge corresponding to long water trees is not released by a low AC voltage, but is released by higher voltage. Based on this characteristic, the method to detect long water trees is suggested here. For convenience, we assume two lengths of water tree in the insulation. The procedure is as follows (Fig.13). (1) A certain DC voltage is applied. (2) A low AC voltage is applied (Va). (3) A high AC voltage is applied (Vb). All of the residual charges corresponding to short water trees are released by a voltage application of Va, but residual charges corresponding to long water trees are not released at this stage. This residualcharge component is released by a voltage application of Vb. Therefore, we can obtain the information on the long water trees, that is, whether there are long water trees or not can be known by estimating the residual-charge component released by applying a high AC voltage. 6. Conclusion We reported the effectiveness of residual-charge measurement for diagnosing water tree deteriorated XLPE cable, the possibility of OSW application to conventional residual-charge measurement and new method for detecting long water trees. Conclusions from this study are as the follows: (1) Residual charge has a good correlation to breakdown voltage up to 22 kv/33 kv XLPE cable. This indicates that the residual-charge measurement is an effective diagnostic method for watertree deteriorated XLPE cable. Va Release of the residual charge corresponding to water trees with short length at Va Residual charge signal Case 1 Case 2 Pattern of AC voltage application Va Vb Release of the residual charge corresponding to water trees with long length at Vb Fig. 13. Pattern of Applied AC Voltage to Detect Highly Deteriorated Region. (2) The possibility of substituting OSW for AC voltage was confirmed. (3) A method was suggested for discriminating the signal corresponding to short and long water trees, of which the latter strongly affects the cable insulation performance. A method for detecting long water trees strongly affecting the dielectric performance of XLPE cable is thought to be the most effective, therefore, the improvement of this method is in progress. Vb References Time No heavily deteriorated region Heavily deteriorated region exists 1) Technical report of JIEE, No. 668, 1998 2) Technical report of CRIEPI W868 3) Uchida, et al. : Annual Conf. of JIEE, No. 186, 1997 7