INFLUENCE OF DISCHARGE PRODUCTS ON THE PROCESS OF IONIZATION AGING OF RUBBER INSULATION

Yu. N. Shumilov

(Presented by the scientific-methodological seminar of the Department of Electrical Insulation and Cable Engineering)

INTRODUCTION

It is known that organic insulation, operating for a long time under conditions of intense ionization, undergoes aging and gradual destruction. The destruction of insulation may be due to the following causes:

a) the chemical action on the insulation of the products of air ionization—ozone and nitrogen oxides, which are strong oxidizing agents;

b) bombardment by ions and electrons released during the ionization of air;

c) the action of high temperature created in the ionization region as a result of a local increase in losses.

In addition to the causes indicated, it has been established that, under prolonged action of discharges inside gas inclusions, secondary ionization products are formed—moisture, acids, products of decomposition of the solid dielectric, etc.—which may exert an indirect influence on the process of ionization in the gas inclusion and on the “lifetime” of the dielectric [1–5]. Rogers [4] and others [2, 3] showed that moisture and acids, depositing on the walls of a gas inclusion, form a thin semiconducting film, which shunts the gas inclusion and causes the ionization to die out. When ionization dies out, the “lifetime” of the dielectric increases and, in this sense, the formation of secondary ionization products is a favorable factor for long-term operation of the insulation. At the same time, there is an opinion that secondary products can diffuse into the solid dielectric and form regions in it with increased conductivity or increased energy losses. This circumstance may cause a reduction in the long-term electrical strength and a shortening of the “lifetime” of the insulation. The hypothesis concerning the influence of ionization products on the electrical characteristics of a dielectric was first expressed by Whitehead [1] and subsequently received partial confirmation in work [6], in which it was shown that, under prolonged action of corona discharges, a noticeable decrease occurs in the specific volume resistivity of polyethylene. However, the question of changes in the electrical strength of dielectrics as a result of diffusion of discharge products into a solid dielectric has practically not been studied.

The present work is devoted to the investigation of processes in gas inclusions as applied to rubber insulation and to the investigation of the influ-

...of the insulation products on the electric strength of high-voltage rubber.

Investigation of Processes in a Gas Inclusion

The investigation of processes inside a gas inclusion was carried out on models with an artificial gas inclusion, shown in Fig. 1 (type A, type B). Type A models were used to investigate the decay of ionization; type B models, to investigate phenomena on the surface of the inclusion walls. The depth of the inclusion in all cases was—

Fig. 1. Sketches of type A models for investigating the decay of ionization and type B models for investigating phenomena on the surface of the inclusion walls:
1—rubber plates (TS-35), 2—electrode holder, 3—high-voltage electrode, 4—low-voltage electrode, 5—mica, 6—glass plate, 7—rims of oil-resistant mastic, 8—transformer oil, 9—air inclusion (for the type A model the inclusion diameter is 1.8–5 mm; for the type B model, 50 mm).

—0.4 mm. An alternating voltage of frequency 50 Hz and 1000 Hz, equal to \(2.5 U_{\mathrm{n}}\) (\(U_{\mathrm{n}}\) is the ionization inception voltage), was applied to the models. At certain time intervals, the ionization intensity \(I\), the specific surface resistance of the walls of the gas inclusion \(\rho_s\), and the hydrogen ion exponent pH were measured for drops of distilled water applied inside the gas inclusion. The hydrogen ion exponent pH served as a measure of the formation of an acidic medium inside the gas inclusion. The ionization intensity was determined according to the method described in [7]; the surface resistance was measured on a galvanometric setup, and pH by the colorimetric method. To determine \(\rho_s\) and pH, type B models were removed from the test stand and disassembled. After the measurements the models were replaced with new ones.

Fig. 2. Dependence of ionization intensity on the duration of voltage exposure:
●●—inclusion diameter 1.8 mm, ○○—inclusion diameter 3.5 mm, ++—inclusion diameter 5 mm

Fig. 2 shows the most characteristic ionization-decay curves for inclusions with diameters of 1.8, 3.5, and 5 mm. It is evident from the figure that the most rapid decay of ionization takes place during the first two hours, after which comparatively slow ionization decay is observed. In inclusions with diamet—

with a diameter of 1.8 mm, complete extinction of ionization occurred within 3–10 h. In inclusions 3.5 mm in diameter, after 50 h of testing complete extinction of ionization took place in only 4 of the 12 models tested. In models with an inclusion diameter of 5 mm, after 10 h the ionization intensity decreased to \(5 \cdot 10^{-12}\) C and practically did not decrease further during the entire test time (about 100 hours).

Table 1 gives the results of measuring \(\rho_s\) and pH for type B models. It is seen from Table 1 that the change in \(\rho_s\) and pH is analogous to the change in ionization intensity. The identity of the dependences \(\lg I = f(\tau)\); \(\lg \rho_s = f(\tau)\) and \(\mathrm{pH} = f(\tau)\) convincingly confirms the extinction mechanism caused by shunting of the gas inclusion.

Table 1

Change in \(\lg \rho_s\) and pH with test time

When the models were dismantled, it was observed that the rubber in the discharge zone acquired a yellow-brown color and that blue crystals formed on the surface of the electrode. Chemical analysis showed that the crystals formed are salts of nitric and sulfuric acids, \(\mathrm{Cu(NO_3)_2}\) and \(\mathrm{CuSO_4}\). The qualitative reactions for determining \(\mathrm{NO_3^-}\) and \(\mathrm{SO_4^{--}}\) ions in solution were, respectively, the reactions with diphenylamine \((\mathrm{C_6H_5})\mathrm{NH}\) and barium chloride \(\mathrm{BaCl_2}\). The formation of nitric acid in a gas discharge in an air medium is well known [9]. The formation of sulfuric acid in gas inclusions is evidently characteristic only of rubber insulation and is apparently associated with the release of sulfur from the rubber under electron (ion) bombardment.

Thus, in rubber insulation, during prolonged exposure to ionization inside gas inclusions, a semiconducting film is formed comparatively rapidly and ionization is extinguished. At a test voltage equal to \(2.5 U_n\), complete extinction of ionization occurs in inclusions 1.8 mm in diameter. In inclusions of larger diameter, ionization may exist for a long time.

Influence of discharge products on the short-term electric strength of rubber insulation

Under prolonged exposure of rubber sheets placed between two glass plates with a gap to discharges, it was found that an acidic medium forms only in the zone of direct action of the discharge. In this connection, treatment of the rubber sheets with ionization products was carried out not indirectly, as recommended in [5], but under the direct action of the discharge. Rubber sheets 0.4 mm thick were placed in the gap between a copper electrode and a glass plate, on the outer surface of which electrodes 15 mm in diameter had been applied, and were held for 30 min at a voltage equal to \(2.5 U_n\) (\(f = 1000\) Hz). The gap between the rubber surface and the glass plate was 0.4 mm. The treatment time and frequency of the test voltage were chosen from conditions of maxi-

of a low yield of the acidic medium and the minimum duration of exposure to the discharge (Table 1). After discharge treatment, the surface of the rubber plates was inspected with a binocular microscope. No signs of destruction of the rubber surface were observed during inspection. Five hours after treatment, the electric strength of the rubber plates \(E_{\mathrm{br}}\) was determined at industrial frequency \((f = 50\ \text{Hz})\). To prevent breakdowns from the edge of the electrode, the samples were immersed in transformer oil with an addition of 20% dibutyl phthalate.

Because of the large scatter in the \(E_{\mathrm{br}}\) values of the rubber plates, tests were carried out on a large number of specimens, and the results obtained were then processed statistically.

Table 2

Distribution series of the electric strength of TS-35 rubber before and after treatment by corona discharges.

Table 2 gives the experimental frequencies for untreated plates (column 2) and for plates treated with discharge products (column 3).

It was found that the distribution of the values of \(E_{\mathrm{br}}\) for the untreated plates obeys the normal law, with a mean value \(\overline{E}_{\mathrm{br}1}\) equal to 46.9 kV and with a standard deviation \(\sigma_1\) equal to 5.47 kV.

\[ f(E_{\mathrm{br}})=\frac{1}{\sigma\sqrt{2\pi}}\cdot e^{-\frac{(E_{\mathrm{br}i}-\overline{E}_{\mathrm{br}})^2}{2\sigma^2}} . \]

A check by Pearson’s criterion [8] shows that the experimental distribution agrees with the normal distribution with a confidence of more than 80%. The confidence limits of \(\overline{E}_{\mathrm{br}1}\) at the 5% significance level, determined as

\[ \overline{E}_{\mathrm{br}1}\pm 1.96\frac{\sigma_1}{\sqrt{n}}, \]

are, respectively, 45.83 kV and 47.97 kV.

The statistics of the experimental series of the treated plates, \(\overline{E}_{\mathrm{br}2}\) and \(\sigma_2\), are respectively 46.2 kV and 5.46 kV. It can be seen that \(\overline{E}_{\mathrm{br}2}\) lies within the confidence limits of the distribution of the electric strength of the first series. Estimation of the variances \(\sigma_1^2\) and \(\sigma_2^2\) by the Romanovsky criterion [8] shows that the difference between the variances of series 1 and 2 is insignificant. Thus, treatment of rubber plates with discharge products does not lead to a reduction in their electric strength.

Conclusions

1. Investigation of ionization processes in rubber on models of a gas inclusion made it possible to detect the formation, within the inclusion, of secondary ionization products—nitric and sulfuric acids and their salts.

2. The formation of secondary ionization products leads to comparatively rapid extinction of ionization and does not lead to a reduction in the electric strength of rubber plates.

3. At a voltage equal to \(2.5 U_n\), \(f = 50\ \text{Hz}\), complete extinction of ionization occurs in inclusions 1.8 mm in diameter within 3–10 hours. In inclusions of larger diameter, discharges of magnitude \((5 \div 10)\cdot 10^{-12}\ \text{C}\) may exist for a long time.

1. In considering the process of ionization aging of rubber insulation, it is necessary to take into account the phenomenon of rapid extinction of ionization.

REFERENCES

1. S. Whitehead. Breakdown of Solid Dielectrics, Gosenergoizdat, 1957.

2. A. M. Thomas, J. I. E. E., vol. 91, pt 11, p. 549.

3. Resistance of sheet insulation to surface discharges. BEARA Report NL/T 379, 1958.
   Translation No. 5238. TsBTI, Scientific Research Institute of the Electrical Industry.

4. Rogers E. S. Proc. Instn. Electr. Engrs, vol. 105A, pp. 621–630, 1958.

5. J. B. Birks, J. G. Schulman. Progress in Dielectrics. Vol. 1, GEI, 1962.

6. V. G. Sotnikov. The influence of partial discharges in an air layer on the electrical resistance of polyethylene. Kabel’naya tekhnika, 1964, No. 33.

7. G. S. Kuchinskii, O. O. Tapupere. Recording ionization characteristics of insulation. Elektrichestvo, 1960, No. 11.

8. A. K. Mitropol’skii. Technique of Statistical Calculations. Fizmatgiz, 1961.

9. A. B. Shekhter. Chemical Reactions in an Electric Discharge. ONTI, L.–M., 1936.