Analysis of Potential Rise on UHV Oil-paper Capacitor Bushing End Shield in the Lightning Impulse Test

Introduction

Grounding is one of top priorities that should be taken into account and plays a vital role in ensuring personnel safety, secure equipment and correct test results. However, attention to grounding is not sufficient and incorrect grounding or forgetting to ground causes hidden troubles, abnormal tests or fatal accidents. To date, there have been many studies on system grounding mode and grounding net, mainly focusing on substation or grounding resistance of overhead line poles or characteristics of diffuse. There is a lack of quantitative analysis of grounding problems in the test. The analysis of ground potential rise (GPR) tends to focus on the impact of grounding resistance and neglect the influence of grounding line’s high-frequency impedance at impulse test conditions. The high-frequency component generated in the impulse test makes ground lines exhibit the characteristics of inductance. As a result, the impulse grounding impedance is more than power frequency grounding impedance, causing GPR and damage to the equipment. However, a mature theory and method of impulse grounding impedance measurement have not been developed at home and abroad until now.

The high voltage bushings tend to be subject to electrical and thermal test for a long time. In recent years, faults in the power transformer have always occurred while bushing breakdowns accounted for 30% or so. Among bushing breakdowns, poor end shield grounding leads to the discharge and further causes oil chromatographic data to exceed the limit, which occupies a large proportion. Hence, it is of great importance to investigate the grounding mode of bushing end shield and end shield to ground potential rise.

Through the impulse test of oil-paper capacitor bushing, it is found that the chromatographic data is abnormal. Therefore, the high-frequency impedance of grounding line is measured and grounding modes of capacitor bushing are compared. Besides, causes of bushing end shield to ground potential rise are analyzed and its value is estimated. The test results can provide the basis for analyzing failure causes, improving the structure of bushing and perfecting the technology of bushing test.

1. Bushing Test

1.1 Test wiring

The high voltage oil-paper capacitor bushing is composed of conductive pole, capacitive layer, porcelain insulator and flange. The aluminum foil and insulation paper are alternately rolled into several capacitive layers to make radial and axial electric fields between conductive pole and flange distribute evenly, which is shown in Fig.1.The outermost capacitive layer is called end shield. The electric outlet wire of end shield is led out by lead column across small bushing. The lead column is insulation against the ground and is connected to the ground via metal grounding cover. The small bushing of end shield is used to measure the capacitance and dielectric loss factor. Hence, it is also called measurement terminal. The rated insulation level of test high voltage bushing is listed in Tab.1.

Tab.1 Rated insulation level of tested bushing

When conducting the power frequency withstand voltage test, both ends of the bushing are placed in the corresponding insulation media and the test voltage is applied on high voltage end of the bushing. The end shield is connected with the flange, which is grounded through the housings of other pieces of equipment. See below for Fig.1.

Fig.1 Grounding mode of end shield in the power frequency test

When performing the lightning impulse test, a resistor (non-inductive resistance with low resistance value) needs to be connected in series outside the measurement terminal for measuring the current and inputting it to the oscilloscope. Under the circumstances, it is necessary to disconnect the measurement terminal from the flange. The grounding mode of end shield adopted in the impulse test is shown in Fig.2. The point B is connected with the grounding point D through the flange according to the proximity principle. The measurement terminal is connected with grounding point C via the resistor (0.1Ω) and grounding copper foil, which is 20m long.

Fig.2 Grounding mode in the impulse test

1.2 Test procedure and abnormal conditions

Prior to the lightning impulse test (including lightning full wave test and lightning chopped wave test), the 1200kV/5min power frequency withstand voltage test, switching impulse test and thermal stability performance test have been performed on the bushing. All the tests are successful and the oil chromatography after the tests are normal. However, after the completion of lightning impulse test, a small quantity of C2H2 is found in the oil during the oil chromatographic analysis. Its volume fraction is 7.6 × 10-6 and the data is specified in Tab.2. Based on that, it is determined that the bushing discharges slightly during the lightning impulse test.

Tab.2 Oil chromatographic data of tested bushing 

The partial discharge of bushing is less than 10pC and nothing abnormal happens. Besides, no discharge or breakdown appears in the later bushing end shield to ground 3kV/1min power frequency withstand voltage test.

In order to determine the source of acetylene and discharge position, the bushing is disassembled. There is nothing abnormal about main insulation of bushing but it is found that the end shield wire is long and the bend of wire is very close to the pedestal. The tinned wire core becomes black and hardens, hinting that the high current passes through end shield wire during the test. The insulation sheath of middle wire is damaged and there are several broken strands, which is shown in Fig.3.

Fig.3 Damage to end shield wire and pedestal

1.3 Analysis of causes of end shield discharge

The GB/T 7252 Guide to the Analysis and the Diagnosis of Gases Dissolved in Transformer Oil presents a three-ratio method. The analysis of causes by means of this method is as follows: to code the ratio of C2H2 volume fraction to C2H4 volume fraction 1, the ratio of CH4 to H2 0 and the ratio of C2H4 to C2H6 2. The causes may be spark discharge between spark discharge or different potentials or floating potentials and oil gap flashover.

The discharge is not likely to occur between the end shied and flange. However, the end shield and flange are grounded separately during the impulse test as per the grounding mode of Fig.2. Under normal conditions, the capacitance between point A and B is end shield to ground capacitance. The BD grounding line is the outer shell of equipment. Its cross-sectional area is large; the distance is short and the impedance value is low. Hence, the point B and D are basically equi-potential. However, the current mainly passes through the branch circuit AC and the potential of point C rises higher than point D. Given that the point A is connected with point C through a long grounding line, the potential of point A must be more than point C. If the potentials of four points are represented by UA, UB, UC and UD, then UA＞UC＞UD ≈ UB. The potential difference UAB between point A and B is bound to increase. When the potential difference increases to a certain extent, the end shield will discharge to the flange.

The most likely cause of discharge is potential rise of end shield. The following presents the detailed analysis of potential rise causes.

2. Analysis of Causes of End Shield Potential Rise and Calculation

2.1 Characteristics of grounding line impedance

The high-frequency component flowing through grounding lines is abundant during the impulse test. So its own inductive features of grounding line will hinder the impulse current from flowing towards the distant end of the grounding body. As a result, the grounding line is not made full use of. The phenomena is called inductive effect. When the inductive effect is obvious, the impulse earthing impedance is more than power frequency grounding impedance. In order to study the high-frequency impedance of earthing wire, the HIOKI3532-50 LCR impedance tester is used to measure the impedance of grounding copper foil used in the impulse test at different frequencies. The method of four-electrode connection is shown in Fig.4.

Fig.4 Schematic diagram of copper foil impedance measurement

The copper foil can be regarded as the conductor, the section of which is rectangular. According to the calculation formula of inductance:

Where:

L-inductance nH

l- the length of conductor cm

b-the width of rectangle section cm

c-thickness cm

μ- relative permeability

δ - skin effect coefficient

The inductance of one single copper foil (1m long, 0.1 wide and 0.5 thick), one single copper foil (0.5m long, 0.1 wide and 0.5 thick) and two copper foils in parallel (0.5m long) is measured. The inductance varies according to the frequency, which is shown in Fig.5.

Fig.5 Relation between the inductance and frequency

From Fig.5, it is found that the inductance value of grounding copper foil is related to its length and cross-sectional area. Between 10kHz and 1MHz, the inductance of 1m copper foil is about 500nH while around 265nH for 0.5m copper foil. The longer the copper foil is, the higher the inductance value is. When the length of copper foil is 0.5m, the inductance of two parallel copper foils is 100nH or so. Hence, an increase in cross-sectional area can decrease the value of inductance effectively, which is in consistent with inductance calculation formula. The skin effect means the impact of frequency. In terms of copper conductor, research indicates that the skin effect coefficient is close to 0.25 at low frequency; the δ is in inverse proportion to the square of frequency at high frequency; the coefficient tends to be zero at very high frequency. The variation of inductance based on the frequency is relatively small, that is, 6%for short wires and 2% for long wires. The test results show that the inductance value of copper foil falls slightly at 100kHz and then remains constant.

Now that the resistance of copper foil is very low and the skin effect is further less than the linear relation between the inductive reactance and frequency, the impedance is mainly dependent on the inductance at high frequency. Based on the real measuring results, the inductance is constant within the range of 10kHz-1MHz. Therefore, the impedance Z is in linear relation to the frequency, which is shown in Fig.6.

Fig.6 Relation between the impedance and frequency

In conclusion, the impedance of grounding copper foil is much more than power frequency impedance. The effective methods of reducing grounding resistance are to shorten the length and increase the section area of grounding lines.

2.2 Estimation of wire potential rise value

The equivalent circuit of impulse test is shown in Fig.7. The Fig.7 can be simplified into Fig.8 because the impedance of four branch circuits are mainly the end shield to ground capacitive reactance.

Fig.7 Equivalent circuit of test connection

Fig.8 Simplified circuit of test connection 

C0-main capacitor of bushing 438pF

Cp-end shield to ground capacitance 5293.5pF

Rm-resistor 0.1Ω

La-equivalent inductance of grounding line AC

Rn-equivalent resistance of grounding line AC

Rb-equivalent resistance between Point B and grounding point D

Lb-equivalent inductance between Point B and grounding point D

RCD-resistance between point C and D

The rise value of point A potential can be expressed as follows:

IAC - the current flowing through the branch circuit AC

ZAC - equivalent impedance of branch circuit AC

ω - angular frequency

Given that the impulse voltage waveform is standard lightning wave, the wave front time 1.2μs can be equivalent to one fourth of sine wave. Its corresponding frequency is 208kHz. The frequency is approximate to fundamental frequency and the effect of other harmonic frequencies is neglected. The equivalent impedance from end shield to grounding point C (including grounding lines) is 12.9Ω or so. The capacitive reactance is as follows:

The voltage and current waveforms measured during the lightning full wave impulse test are shown in Fig.9. The channel 1 refers to the voltage lessened from low voltage arm voltage of voltage regulator (the voltage ratio is 2550:1) as per ratio (attenuation ratio is 100:1) and input into the oscilloscope. Each grid of vertical axis means 5kV voltage and the applied impulse voltage is 1275kV; the channel 2 refers to the voltage lessened from the voltage of resistor (0.1Ω) as per the ratio (attenuation ratio is 100:1) and then input into the oscilloscope. Its each grid means 2V voltage and its current is 2000A.

Fig.9 Voltage and current waveforms in the negative lightning impulse test

According to the conversion of current waveform, the peak value of current flowing through the branch circuit AC is 1200A or so during the full wave impulse test. Then the potential difference between point A and C is calculated: 

UAC=IACZAC=15.4kV            (4)

The current flowing through Cp is listed below:

ICP=UAC/XC=106.8A            (5)

The current flowing through the main capacitor should be the sum of two branch circuits, that is, 1306.8A.

It is assumed that the lightning impulse voltage peak value Up is 2400kV, then the current passing through the main capacitor is calculated below:

IC0=(Up-UAC)ωC0=1365A       (6)

Therefore, the calculated value of current flowing through the main capacitor during the lightning full wave test is basically consistent with measuring results.

As per the same method, the current peak value is 1600A and the chopped frequency is about 416kHz during the negative chopped test, which is shown in Fig.10. The equivalent impedance of branch circuit AC is 25.6Ω; end shield to ground capacitive reactance is 74.3Ω; the potential difference between point A and C is calculated at 41kV. Under normal conditions, the end shield to ground current is 567.3A. However, the current flowing through main capacitor should be 3112.8A, which is 945A more than the sum of two circuit current. Because the end shield outlet wires are broken and close to inner wall, the potential rise makes defective insulation layer and oil gap breakdown and discharge during the chopped wave test.

Fig.10 Voltage and current waveforms in the lightning chopped wave test

The same method can be used to estimate relevant parameters for switching impulse test if conditions are the same. The calculated results under different impulse waveforms are listed in Tab.3.

Tab.3 Results at different impulse waveforms

To conclude, during the lightning impulse chopped wave test, owning to high equivalent frequency and high impedance of grounding line, the potential of end shield wire reaches 41kV and the flowing current is high. That is basically consistent with bushing breakdown.

2.3 Insulation test of end shield wire

From the above calculation, we know that the end shield potential reaches 41kV. In order to analyze whether the end shield wire breakdowns at 41kV, insulation test is performed on the end shield. The wires of 1100kV oil-paper bushing end shield is shown in Fig.11.

Fig.11 End shield of bushing 

Five end shield outlet wires with the same form are selected to conduct the power frequency withstand voltage test. The test results are shown in Tab.4.

Tab.4 Power frequency withstand voltage test results of the end shield outlet wire

The minimum voltage of power frequency breakdown is 17kV and the average voltage is 19.2 according to test results. In light of 1.5 times lightning impulse coefficient, the lightning impulse breakdown voltage is 28.8kV. However, in the chopped wave impulse test, the maximum potential of end shield can reach 41kV, 2.14 times of power frequency breakdown voltage. Hence, it can speculated that the wire potential rise causes the insulation damage. 

3. Improvement Measures

The root cause of end shield discharge is an increase in potential difference between the end shield and flange caused by end shield potential rise and two-point grounding. Hence, improvement measures should be taken from two aspects: grounding impedance and grounding mode. On the one hand, the length of grounding line needs to be shortened and the cross-sectional area be enlarged; on the other hand, one-point grounding instead of two-point grounding is recommended. It is wise to adopt the grounding mode shown in Fig.12 if the resistor must be used to monitor the current. The flange is connected to the back of resistor via short wires and then to the grounding point C. By doing so, the interference in measurement of current can be avoid and the potential difference between point A and B can be reduced.

Fig.12 Improved grounding mode of impulse test

4. Conclusions

1) The impedance of grounding line is much more than power frequency impedance at high frequency; the high-frequency impedance is mainly determined by the inductance. The longer the wire is, the higher the inductance is. The inductance is in inverse proportion to the cross-sectional area.

2) The end shield wire to ground potential is up to 41kV during the lightning impulse chopped wave test. At that time, the end shield wire generates spark discharge to the ground, resulting in abnormal oil chromatographic data.

3) Keep an eye on the high-frequency impedance of grounding line which may cause earthing potential rise during the impulse test. The cross-sectional area of wire should be enlarged; the length of grounding lines should be shortened and one-point grounding method is recommended.