Study on the dc flashover performance of standard suspension insulator with ring-shaped non-uniform pollution

Under dc electric field, pollution may be more easily deposited around insulator string caps and pins, distributing nonuniformly on its inner and outer parts, and presenting ring-like shape. dc Flashover tests of 7-unit XHP-160 insulator string were carried out. During the tests, the ring-shaped non-uniform pollution was simulated using solid-layer method by changing the ratio I/O of inner-to-outer surface salt deposit density (SDD) and radius (r) of the heavily polluted area (the inner part). On the basis of the test data, the effects of ring-shaped non-uniform pollution on dc flashover characteristics were studied. Then, the relative variation trend of flashover voltage, surface pollution layer resistance, and leakage current were discussed. Research results show that the ratio I/O affects the dc flashover performance of the insulator string. The flashover voltage increases with the increase of I/O. With the increase of the radius r, the flashover voltage will increase at first and then decrease. Under ringshaped non-uniform pollution, surface pollution layer resistance of insulator will increase, which results in lower leakage current and higher flashover voltage. The results will be of certain value in providing references for outdoor insulation selection.


Introduction
It is known that the success rate of transmission line lightning flashover re-closing is higher than that of pollution flashover, and so the loss caused by pollution flashover fault of external insulation is far greater than the lightning fault [1][2][3].In recent years, along with the rapid development of industry and economy, the air quality is getting worst, making the pollution flashover still a threat to electric power system security [3].
The non-uniformity of the natural contamination can be divided into four ways: between the top and bottom surfaces of insulator, along the periphery, along the insulator length, and among the different insulator units of the same string [3][4][5].To make the artificial test more comparable with the actual situation, relevant institutes have made a lot of studies on the flashover performance and mechanism under non-uniform pollution flashover.
For instance, as is presented in the literature [6], insulator nonuniform pollution between top and bottom surfaces will affect flashover voltage, and this effect is associated with insulator shed structure.The literature [7] indicated that the ratio of non-uniform pollution from top to bottom surfaces (T/B) is in the range of 1/5-1/10.The literature [8] shows that the flashover voltage increases by ∼30 and 50%, when T/B changes from 1/1 to 1/5 and 1/10.Benchmarked against the salt deposit density (SDD) of bottom surface of insulator, Electric Power Research Institute (EPRI) [9] raised a formula for the correction of dc flashover voltage under non-uniform pollution, and got that the correction coefficient (C) was in the range of 0.29-0.47.In the literature [10], another research was done to study on the influence of non-uniform pollution distribution on ac flashover voltage of standard suspension insulators, which found that the formula of EPRI is also applicable in ac case and the value of C was obtained as 0.31.Moreover, some researches about other kinds of insulators' non-uniform pollution were reported recently.For example, in [11], the researchers studied the fan-shaped non-uniform pollution between windward and leeward sides of insulator string and its effects on dc flashover performance.The results indicated that flashover voltage may decrease by 28% under certain situation [11].Researchers in [12] investigated the dielectric performance of a plane model using wavelet transform with respect to the position and width of pollution layer.Another research in [13] presented the effects of discontinuous non-uniform pollution on the flashover of polluted insulator under impulse voltage.In [14], the estimation of the effective pollution thickness was proposed when studying the effects of pollution layer linear non-uniformity on insulator flashover.Moreover in [15], effects of pollution distribution class on insulators flashover were studied, which leads to a conclusion that the insulator with non-uniform transverse distribution pollution will have a higher withstand voltage than those with uniform contamination.
So far, works on T/B non-uniform pollution has been done around the world and the related results are applied in insulation coordination by designers.However, in operating lines, especially under dc electric field and variable wind directions, contaminants may get more easily deposit around insulator string's caps and pins, forming like a ring-shaped non-uniform distribution.As shown in Fig. 1, the observations of Figs.1a-c are obtained from the 1 year non-energised insulator strings suspended in Xuefeng Mountain Test Station, where it features a continental climate.The observations of Fig. 1d are obtained from the 18 months nonenergised insulator string on the rooftop of the high-voltage laboratory in urban area.The observations of Figs.1e-f are obtained from the 1 year dc-energised insulator string located on the 500 kV dc Jinsu line, where also features a continental climate.It may have some effects on its flashover voltage.So, the research on it is referable and valuable not only for the further insight of insulator flashover performance, but also for the better design of outdoor insulation coordination.
2 Sample, experimental setups, and procedure configuration height, L is the leakage distance, and D is the diameter of insulators.

Experimental setups
The tests were carried out in the multi-function artificial climate chamber.The artificial climate chamber, which is with a diameter of 7.8 m and a height of 11.6 m, can simulate complex atmospheric environments such as fog, rain, ice, and high altitude [16][17][18][19][20].The dc power supply is a 600 kV/0.5A cascade rectifier circuit controlled by the thyristor voltage-current feedback system which ensures that the voltage ripple factor is <3.0%when the load current is 0.5 A. The dynamic voltage drop of the power supply is no more than 5%, and the relative voltage overshoot due to load release is <8%.The test power supply satisfies the requirements recommended by [16][17][18][19][20].
The test circuit is shown in Fig. 2. In the circuit, the 10 kV/AC source offers a power rating of 300 kW, S 1 and S 2 are the front switch and post-switch, respectively, B represents a power transformer of 150 kV/900 kVA, C 1 , C 1 ′, C 2 , and C 2 ′ are capacitors (the capacitance 0.613 μF, voltage rating 213 kV), D is highvoltage silicon stack, Silicon Controlled Rectifier (SCR) is the silicon control elements, R 0 is a current limiting resistor (10 kΩ), r is a current sampling resistor (1 Ω), G is the discharge tube used for protecting the acquisition system (voltage rating 5 V), H is a 330 kV wall bushing, R 1 and R 2 are the resistors of the resistance divider F (voltage division ratio 15:1, input resistance 1800 MΩ, and voltage rating 600 kV), and the accuracy of F is 0.5%.E is the artificial climate chamber.

Test procedure 2.3.1 Preparation:
Before the tests, all the samples were carefully cleaned by Na 2 PO 3 solution, so that all traces of dirt and grease were removed.Then, the samples were thoroughly rinsed with tap water, and let to dry naturally indoor to avoid dust or other pollution.

Polluting:
The soluble contaminants were calculated by SDD.SDD represents the weight of sodium chloride (NaCl) per unit area of insulator, in 'mg/cm 2 '.Moreover, non-soluble material Kieselguhr was calculated by non-soluble deposit density (NSDD).SDD was selected for 0.06, 0.10, and 0.25 mg/cm 2 to represent three different levels of pollutions.The ratio of NSDD to SDD was 6 in all the tests.The contamination ratio of the inner-to-outer surface of the ring-shaped of the 7-unit XHP-160 insulator strings was generally selected for 1:1, 3:1, 8:1, and 15:1 to represent four different degrees of pollutions.
The method of simulating ring-shaped non-uniform pollution is described as below.Through observing insulator natural contamination, it was found that the inner position of surface was more seriously polluted as shown in Fig. 1.Thus, the diagram of ring-shaped non-uniform pollution is shown in Fig. 3b, where region 1 is the steel cap of the insulator, region 2 is the heavily polluted area, named the inner part, and region 3 is the lightly polluted area, named the outer part.To enable artificial nonuniform pollution, the edges on the bottom surface of the insulator were regarded as the boundary of selecting the radius (r) of the inner part, as shown in Fig. 3c, and thus the radii (r) of the inner part were 8, 11, and 12.5 cm, respectively.The ring shape on the top surface and the bottom surface was divided by the same r.
The insulators were polluted using solid-layer method by pasting.The SDD of the inner part (SDD I ) and the outer part (SDD O ), and the average SDD of the porcelain insulators can be expressed as (2), and the data of SDD I and SDD O in this paper are shown in Table 2 where S I , S O are areas of the inner part and outer part of the surface, I/O represents the ratio of inner-to-outer surface SDD.

Flashover test:
The predicted flashover voltage level should be obtained for the first up-and-down evaluation.First, a small voltage was applied on the insulator string, and then the steam fog, for which the rate was 0.05 ± 0.01 kg/h m 3 , was  supplied.The degree of wetness of the insulator surface was judged by observing the surface water film situation as well as the variation of recorded leakage current waveform.When the surface water films were formed and connected with each other, or the leakage current reached the maximum value and then started decreasing, the insulator pollution surface was treated as wholly wetted.Then, the voltage was raised until flashover.The final voltage value was taken as the expected flashover voltage U 50 .
In the tests, up-and-down method was adopted [16][17][18][19][20].Each contaminated sample was subjected to at least ten 'valid' individual tests.The applied voltage level in each test was varied according to the up-and-down method.The voltage step was ∼5% of the expected U 50 .The first 'valid' individual test was selected as being the first one that yields a result different from the preceding ones.Only the individual test and at least nine following individual tests were taken as useful tests to determine U 50 .The U 50 and standard deviation (σ) are calculated as follows: where U i is an applied voltage level, n i is the number of tests carried out at the same applied voltage U i , and N is the total number of 'valid' tests.
The test procedure of the up-and-down method has been summarised as shown in Fig. 4.

dc Flashover test results
Following the procedures above, dc flashover tests of 7-unit XHP-160 insulator strings polluted under different SDDs, r, and ratio (I/O) were carried out.The results are shown in Table 3.
According to the test results, conclusions can be made as follows: i.The standard deviations of these results are all <7%, which means that the dispersion of the test results is slight.ii.The flashover voltage of insulator string decreases with the increase of SDD.For example, when I/O is 3/1, r is 11 cm, and the values of SDD is 0.06, 0.10, and 0.25 mg/cm 2 , the corresponding U 50 is 106.1, 97.1, and 85.2 kV, respectively.From the data above, it can be seen that the voltage decreases by 8.5 and 19.7% when the SDD increases from 0.06 to 0.10 and 0.25 mg/cm 2 .iii.The ratio (I/O) of the inner-to-outer surface SDD has significant influence on U 50 .Moreover, the flashover voltage U 50 of the insulator string increases with the increase of I/O.For example, the U 50 is 85.4,97.1, 100.5, and 106.9 kV when SDD is 0.10 mg/cm 2 , r is 11 cm, and I/O changes in the range of 1/1, 3/1, 8/1-15/1.The data indicate that there is an increase of 13.7, 17.7, and 25.2% when the ratio I/O increases from 1/1-3/1, 8/1, and 15/1.iv.With the radius (r) of heavily polluted area increases, the flashover voltage of the insulator strings may experience an increasing trend at first and then decrease.For example, when SDD is 0.1 mg/cm 2 , the ratio I/O is 8/1, and the value of r is 8, 11, and 12.5 cm, the corresponding U 50 is 94.8, 100.5, and 90.8 kV, respectively.Compared to the value of r = 11 cm that the voltage decreases by 5.7 and 9.7% when the r is 8 and 12.5 cm.The disparity between the data was slight, but the trend does exist according to repeated tests.

Relationship between U 50 and SDD
A large number of experiment results indicate that the relationship between U 50 and SDD can be expressed as follows [21,22]: where a is a constant related to insulator material, structure, power type, pressure etc.Moreover, n is a characteristic index which reflects the influence of SDD on U 50 .This formula is suitable for predicting the U 50 of insulator string with a known SDD.However, there is no idea that whether the formula is also proper for calculating the U 50 under ring-shaped non-uniform pollution.According to Table 3, the relationship between U 50 and SDD is shown in Fig. 5, where the curves are all fitted by exponential function as is expressed in (5).The fitting results were presented in Table 4.
It can be seen from Fig. 5 and Table 4 that:  i.The exponential curve fitting between U 50 and SDD is good, because all the fitting degrees R 2 are nearly 1. ii.The results from the curve shapes in Fig. 5 show that the influence of SDD on U 50 is similar under different I/O and r.
In Table 4, the value of n which is in the range of 0.13-0.23 is relatively changeless under any conditions.iii.Taking r = 8 cm as an example, when I/O increases from

Effects on critical leakage current
The ratio (I/O) of insulator string affects its pollution flashover voltage.For example, when r = 11 cm, the relationship between I/O and U 50 is shown in Fig. 6.
It can be seen from Fig. 6 that: the values of U 50 increase with the increase of I/O.This result may be associated with critical leakage current.
Leakage current signal contains a wealth of information in reflecting the operational status of the insulators.In this paper, the leakage current just before flashover, namely the critical leakage current I CR , was selected as the characteristic parameter of discharge process.Fig. 7 shows the leakage current waveform when the insulator flashover occurs, generally, the peak value before flashover is defined as critical leakage current I CR .
During the test, I CR of each flashover test were recorded, and the average values corresponding to each pollution condition were obtained and shown in Fig. 8.
Fig. 8 shows that, in a certain SDD, the insulator surface critical leakage current decreases with increase of I/O.For example, when SDD=0.06 mg/cm 2 , the critical leakage current values are 0.27, 0.20, 0.14, and 0.10 A when the I/O ratio increases from 1/1, 3/1, 8/1-15/1.This change is due to the non-uniform distribution of the insulator surface contamination, which causes the variation of its surface conductivity.

Effects on surface pollution layer resistance
The radius (r) of heavily polluted area of insulator affects its flashover voltage.For example, when I/O = 8/1, the relationship between them is shown in Fig. 9.
It can be seen from Fig. 9 that: with the increase of r, the flashover voltage of insulator string will increase at first and then decrease.This result is inseparable with the surface pollution layer resistance.
During the artificial tests, the samples were polluted by NaCl uniformly in each section, so the surface pollution layer conductivity (γ) is directly proportional to the SDD when they are at the same temperature and saturated enough.Moreover, (2) shows that when I/O > 1, SDDO < SDD, SDDI > SDD, namely an increase of I/O will cause the surface layer conductivity of the outer part to decrease; in return, the contamination layer resistance will increase; for the inner part of the surface the situation will be opposite, its contamination layer resistance will decrease.On the basis of the series resistance model of insulator pollution flashover, under ring-shaped non-uniform pollution the surface pollution layer of insulator string can be equivalent to (Fig. 10), where L Ii is the creepage distance of inner part of the ith insulator string unit under a certain r (red lines in Fig. 3a); L Oi is the corresponding outer part creepage distance (yellow lines in Fig. 3a); and a is the 'effective insulator diameter' which can be calculated as follows [23][24][25][26]: where L c is the creepage distance of insulator string, f is the shape coefficient or form factor of the insulator, which can be expressed by So, the equivalent surface pollution layer resistance (R eq_non ) under ring-shaped non-uniform pollution can be obtained where L O and L I are the creepage distances of outer part and inner part, respectively, and L c = L O + L I ; γ O and γ I are the surface pollution layer conductivity of outer part and inner part, respectively.Given the surface pollution layer conductivity (γ) proportional to SDD, and by ( 2) and ( 8), the relationship of the insulator surface contamination resistance under uniform pollution (R eq_uni ) and under non-uniform pollution (R eq_non ) can be expressed by ( 9) According to (9) it can be seen that when I/O > 1, η is always larger than 1, meaning that ring-shaped non-uniform pollution will cause pollution layer resistance to increase, thus impact flashover voltage.The bigger the inner part radius r is, the larger L I /L O under It can be seen from Fig. 11 that: with the increase of L I /L C , the value of η will increase at first and then decrease.This explains that when r increases, the surface pollution layer resistance will change, which affects flashover voltage.Also with the increase of I/O, η will keep increasing.This is consistent with the experimental data, because the leakage current decreases with the increment of pollution layer resistance, making the dry band less easily form on insulator surface than before, and a higher voltage will be needed to maintain the ignition of partial arcs and flashover.

Conclusions
In this paper, the flashover performance of 7-unit XHP-160 insulator strings polluted by different SDD, I/O, and r was studied.Through analysis the following conclusions can be obtained:

Fig. 4
Fig. 4 Flow chart for the test procedure

Fig. 6 Fig. 7
Fig. 6 Relationship between U 50 and I/O (r = 11 cm) Fig. 7 Waveform of leakage current along the surface of insulator a certain L C (total creepage distance).η Values under different I/O and L I /L O are shown in Fig. 11.
i.The flashover voltage U 50 of the insulator string increases with the increase of I/O, and may increase by nearly 36% under certain condition.With the increase of radius (r) of heavily polluted area, the flashover voltage will experience an increasing trend at first and then decrease.ii.The variation of insulator surface pollution layer resistance well explains the effects of different ring-shaped nonuniformity conditions (I/O and r) on insulator pollution flashover performances.With the increase of r, L I /L C will increase, causing the change of surface pollution layer resistance, thus affecting flashover voltage.Also with the increase of I/O, η will keep increasing, which rises surface pollution layer resistance, making the dry band more hardly form on the pollution layer surface, and a higher voltage will be needed for flashover.iii.It is obtained through the systematical research of this paper that insulator flashover voltage will increase if under ringshaped non-uniform pollution.Therefore, the current external insulation configuration of transmission line has sufficient safety margin according to existing standards.However, if insulators in certain area are ring-shaped non-uniform polluted all the time, it is recommended to properly reduce insulator units to save insulation space.

Fig. 8 Fig. 9 Fig. 10 Fig. 11
Fig. 8 Relationship between I CR and I/O of 7-unit XHP-160 insulator string (r = 11 cm) The samples were XHP-160 standard suspension insulators.The technical parameters are shown in Table 1, in which H is the High Volt., 2018, Vol. 3 Iss.2, pp.133-139 This is an open access article published by the IET and CEPRI under the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/)

Table 1
Parameters of the insulator sample Fig. 2 Schematic diagram of the dc test circuit

Table 4
Fitting results of U 50 -SDD under non-uniform pollution