Garton Effect and Its Influence on High-Voltage Dielectric Loss Measurement for Capacitive Equipment WANG Shao-hua1, FANG Yu-qun2 (1 Zhejiang Electric Power Test and Research Institute, Hangzhou 310014, China; 2 Jinhua Electric Power Bureau, Jinhua 321017, China) Abstract: The high-voltage dielectric loss (tanδ) measurement is an effective method for the evaluation of insulation condition of equipment. Introduction was made to Garton effect s influence on the dielectric loss measurement and its formation mechanism. On this basis, combined with the measuring examples, this paper proposed the judgment of Garton effect and treatment methods. It is pointed out that during the high-voltage dielectric loss test, the voltage should be applied evenly, so that the tanδ-u curve will be smooth. After the step-up test, a step-down test should also be conducted. The defect type of equipment insulation can be evaluated through the analysis of the change rule of tanδ-u curve, which helps to ensure the security and economical efficiency of equipment operation. Key words: Garton Effect; capacitive equipment; dielectric loss 0 Introductions DL / T 596-2005 "Preventive Test Procedures for Electrical Equipment" notes that dielectric loss factor (tanδ) and capacitance measurement is a very important content for testing equipment insulation materials, and the value of dielectric loss plays an important role on determining equipment insulation situation[1]. Currently, dielectric loss test under 10 kv is mainly applied to determine whether insulation is with the presence of moisture, oil, dirt or impregnation, deterioration and other defects in field trials. However, the test voltage of 10 kv is much lower than the operating voltage of the equipment, and does not reflect the true condition of the equipment operation. When the equipment is with the presence of moisture, partial discharge or conductive impurities and other defects, its tanδ value is greatly influenced by the test voltage (U) value. By measuring the tanδ and test voltage curve, insulation defects can be more effectively diagnosed [2-5]. "25 Key Requirements for Major Electricity Production Accidents Prevention" (State Power [2000] No. 589) provides that: "For 220 kv and above voltage transformer, dielectric loss test should be carried out under high voltage "[6]. Q / GDW 168-2008 "Power Transmission Equipment State Maintenance Testing Procedures" provides that for current transformers, high voltage capacitive bushing,
circuit breaker shunt capacitors and other equipments, if the dielectric loss factor under 10 kv exceeds the value of attention, it is necessary to do rated voltage dielectric loss measurement (diagnostic test), measuring tanδ-u curve for reference [7]. This paper aims to expound Garton Effect s influence on the dielectric loss measurement and its formation mechanism, on this basis, combined with measuring examples, this paper proposed the judgment of Garton effect and treatment methods, and with view to provide a reference for the analysis of high voltage dielectric loss test data. 1 Garton Effect and Its Formation Mechanism In 1940, Professor M.Garton found that in the medium insulation containing paper (or plastic and oil), the measurement value of tanδ at the lower voltage may be 1 ~ 10 times higher that of the measurement value at high voltage. This phenomenon is called Garton Effect.[8]. The reason of Garton Effect is that the movement of colloidal charged particles is blocked by the paper fibers in the oil under the effect of electric field. This resistance decreases with the increase of electric field intensity. At a lower voltage, colloidal particles are free in the insulating medium, and polarization loss is relatively large, as shown in Figure 1 a); and at higher voltages, these colloidal particles in a strong electric field distribute on both electrodes, the impurities relatively reduce, and then polarization loss is relatively small, as shown in Figure 1 b). Thus, tanδ value for the oil with colloidal charged particles decreases much with increasing electric field strength. For the oil containing ionic charged particles, the resistance of the paper fibers by electric field is not obvious, and therefore tanδ value varies smaller with the electric field strength.
2 Measuring Examples Example 1: A 220 kv paper capacity-type current transformer, the basic parameters: Model LB-220; rated primary current of 2 600 A; rated secondary current of 5 A. Remove from the field and placed long-term on laboratory. Its tanδ test data is shown in Table 1. From the data in table 1, relation curve tan δ value changing with the test voltage U is shown in figure 2. Data analysis for Table 1: for already outage, long-standing current transformer (CT), impurities and moisture distribution situation in the oil is different with the CT under operating conditions. For the CT under operating condition, because of the effect of electric field, impurities and moisture adheres to the surface of the capacitive screen as well the inner wall of bushing between capacitive screens; and for the long-standing CT, impurities and moisture are suspended. When the laboratory test dielectric loss, ions space charge polarizes severely and leads to relatively high value of tan δ. At the same time, from Figure 2, bottom half of tan δ -U curve has downward
trend. Analysis of the reasons: the main insulation of CT LB-220 is a capacitive insulation which made of alternating layers of multi-layer cable paper and aluminum foil or a thin semiconductor paper; in order to make the inner insulation dries easily, evenly spaced holes are punched on aluminum foil. If the CT runs longer, holes on the aluminum foil may be blocked by impurities, when drying the CT, outer insulation is easy to exclude moisture, but inner insulation not, which causes the insulation is not completely dried. When a main screen with not thoroughly dried insulation which has the presence of moisture and impurities, as the voltage increases, the ions velocity will speed up, power loss reduces, making tan δ decreased with the increased voltage. Moreover, in the case of continuous testing, the heat generating will also reduce polarization loss of ionic impurities. Example 2: A generator stator windings, the basic parameters: the rated capacity of 659.34MW; stator voltage of 20kV; stator current of 21149A; power factor of 0.9; single relative capacity of 0.33 ~ 0.34μF. To measure the dielectric loss and capacitance under the sequence of step-up and then step-down, test results are shown in Table 2. From the data in Table 2, relation curve of tan δ value with the change of test voltage U shown in Figure 3. As shown in Table 2 and Figure 3, the tan δ value of the generator stator windings at rated voltage of 20kV was 2.507 percent, which is in a better level.
For analytical judgment to tan δ, should not only pay attention to its absolute value, but also to contrast with the last measurement, as well to the amount of relative change and the relative change amount at different voltages. If the tan δ value at low voltage is small, but varies largely at different high voltages, also shows that the equipment insulation is in poor condition during operation [9]. Figure 4 shows the relation curve for tan δ value of capacitive equipment varies with the test voltage U under different typical defects [10-11]. From the data in Table 2, tan δ increment [tan δ (U n) - tan δ (2kV)] is 0.819%, less than 1.500%, indicating that the overall insulation in good level. When doing dielectric loss test on capacitive equipment, if the tan δ value at a low voltage exceeds the attention value specified in procedure, the equipment should not be easily determined to be ineligible. Taking into account the possible existing Garton Effect, the test voltage should be increased and implementing high-pressure dielectric loss test. A comprehensive judgment to the equipment should be made based on the tan δ - U curve trends. 3 Conclusions During the high-voltage dielectric loss test, the voltage should be applied evenly,
so that the tanδ-u curve will be smooth. After the step-up test, a step-down test should also be conducted; the defect type of equipment insulation can be evaluated through a complete tanδ-u curve. Correct understanding to Garton Effect and its influence on dielectric loss factor tan δ for capacitive equipment, rational analysis to tan δ data and making right judgment and treatment of equipment performance, is of great significance in shortening equipment maintenance time and ensuring the security and economical efficiency of equipment operation. References [1] DL / T596-200 "Preventive Test Procedures for Electrical Equipment" [S]. [2] Huang Min Discussion on High Voltage Dielectric Loss Test for Capacitive Current Transformer [J] Shanghai Electric Power, 2008 (4): 406-409. [3] Li Fan, Yan Chunyu, Chen Xiangyu, Research on High-voltage Current Transformer Dielectric Loss Measurement Method [J] North China Electric Power Technology, 2009 (5): 14-16. 2009 (5): 14-16. [4] Chang Meisheng, Hao Lijun, Analysis of Capacitance and Dielectric Loss Test on Capacitor Voltage Transformer [J]. Journal of Electric Power, 2009,24 (1): 28-30. [5] Li Juanrong, Weng Wei. Measurement on CVT Dielectric Loss Factor and Capacitance at Rated Operating Voltage [J] Northwest Electric Power Technology, 2006 (4): 28-29. [6] State Power [2000] No. 589 [S]. 25 Key Requirements for Major Electricity Production Accidents Prevention [7] Q/GDW168-2008. Testing Procedures for Transmission and Transformation Equipment Maintenance [S]. [8] Liu Chao, Su Minghong, Analysis on Dielectric Loss Test for Capacitive Current Transformer at Rated Voltage [J] Sichuan Electric Power Technology, 2010,33 (5): 44-46. [9] DL/T492-2009. Aging Qualification Guidelines for Generator Stator Winding Epoxy Mica Insulation [S]. [10] Yan Chunyu, Chen Zhiyong, Gao Jun. Dielectric Loss Measurement on Current Transformer at High Voltage [J] High-voltage Electrical Appliances, 2009,45 (2): 87-89. [11] Chen Zhiyong,Yan Chunyu, Zhang Jianzhong. Measurement on Current Transformer Dielectric Loss Factor tanδ at High Voltage [J] Hebei Electric Power Technology, 2005, 24 (5): 33-34. Revised draft date: July 15 th, 2011